Mississippi v. Environmental Protection Agency

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued November 16, 2012 Decided July 23, 2013

No. 08-1200

STATE OF MISSISSIPPI,
PETITIONER

v.

ENVIRONMENTAL PROTECTION AGENCY,
RESPONDENT

COUNTY OF NASSAU, ET AL.,
INTERVENORS

Consolidated with 08-1202, 08-1203, 08-1204, 08-1206

On Petition for Review of a Final Rule Issued
by the United States Environmental Protection Agency

F. William Brownell argued the cause for State of
Mississippi and Industry Petitioners. With him on the briefs
were Allison D. Wood, Lucinda Minton Langworthy, Harold
E. Pizzetta, III, Special Assistant Attorney General, Office of
the Attorney General for the State of Mississippi, Robert R.
Gasaway, Jeffrey Bossert Clark, and William H. Burgess.

David S. Baron argued the cause for Environmental
Petitioners. With him on the briefs was Seth L. Johnson.
 2
Michael J. Myers, Assistant Attorney General, Office of
the Attorney General for the State of New York, argued the
cause for State Petitioners. With him on the briefs were Eric
T. Schneiderman, Attorney General, Barbara D. Underwood,
Solicitor General, Morgan A. Costello, Assistant Attorney
General, Kamala D. Harris, Attorney General, Office of the
Attorney General for the State of California, Nicholas Stern,
Deputy Attorney General, George Jepsen, Attorney General,
Office of the Attorney General for the State of Connecticut,
Kimberly P. Massicotte and Scott Koshwitz, Assistant
Attorneys General, Joseph R. Biden, III, Attorney General,
Office of the Attorney General for the State of Delaware,
Valerie M. Satterfield, Deputy Attorney General, Lisa
Madigan, Attorney General, Office of the Attorney General
for the State of Illinois, Gerald T. Karr, Senior Assistant
Attorney General, William J. Schneider, Attorney General,
Office of the Attorney General for the State of Maine, Gerald
D. Reid, Assistant Attorney General, Douglas F. Gansler,
Attorney General, Office of the Attorney General for the State
of Maryland, Roberta R. James, Assistant Attorney General,
Martha Coakley, Attorney General, Office of the Attorney
General for the Commonwealth of Massachusetts, Carol
Iancu, Assistant Attorney General, Michael A. Delaney,
Attorney General, Office of the Attorney General for the State
of New Hampshire, K. Allen Brooks, Assistant Attorney
General, Gary K. King, Attorney General, Office of the
Attorney General for the State of New Mexico, Stephen R.
Farris, Assistant Attorney General, Ellen F. Rosenblum,
Attorney General, Office of the Attorney General for the State
of Oregon, Paul S. Logan, Assistant Attorney-in-Charge,
Peter Kilmartin, Attorney General, Office of the Attorney
General for the State of Rhode Island, Gregory S. Schultz,
Special Assistant Attorney General, Irvin B. Nathan, Attorney
General, Office of the Attorney General for the District of
Columbia, Amy E. McDonnell, Deputy General Counsel, and
 3
Christopher King. William L. Pardee, Assistant Attorney
General, Office of the Attorney General for the
Commonwealth of Massachusetts at the time the brief was
filed, Tricia K. Jedele, Special Assistant Attorney General,
Office of the Attorney General for the State of Rhode at the
time the brief was filed, Maureen Smith, Senior Assistant
Attorney General, Office of the Attorney General for the State
of Rhode Island, Donna M. Murasky, Deputy Solicitor,
Office of the Attorney General for the District of Columbia,
and Katherine Kennedy entered appearances.

Richard A. Wegman and Harold G. Bailey, Jr. were on
the brief for amicus curiae Province of Ontario in support of
petitioners State of New York, et al. and petitioners American
Lung Association, et al.

Madeline Fleisher, Attorney, and David J. Kaplan,
Senior Attorney, U.S. Department of Justice, argued the
causes for respondent. With them on the brief was John
Hannon, Attorney, U.S. Environmental Protection Agency.

David S. Baron and Seth L. Johnson were on the brief for
Environmental Intervenors.

F. William Brownell, Allison D. Wood, Lucinda Minton
Langworthy, Robert R. Gasaway, Jeffrey Bossert Clark, and
William H. Burgess were on the brief for industry intervenor-
respondents Ozone NAAQS Litigation Group, et al. in
support of respondent. Duane J. Desiderio entered an
appearance.

Before: TATEL, BROWN, and GRIFFITH, Circuit Judges.

Opinion for the Court filed PER CURIAM.
 4
PER CURIAM: In this opinion, we consider several
challenges to the Environmental Protection Agency’s most
recent revisions to the primary and secondary National
Ambient Air Quality Standards for ozone. For the reasons
given below, we deny the petitions, except with respect to the
secondary ozone standard, which we remand for
reconsideration.

I.
The Clean Air Act directs EPA to establish and
periodically review and revise primary and secondary
National Ambient Air Quality Standards (“NAAQS”) for
certain pollutants the “emissions of which . . . cause or
contribute to air pollution which may reasonably be
anticipated to endanger public health or welfare.” 42 U.S.C.
§ 7408(a)(1)(A). Under section 109(b)(1), primary NAAQS
are to be set at levels “the attainment and maintenance of
which in the judgment of the Administrator, . . . allowing an
adequate margin of safety, are requisite to protect the public
health.” Id. § 7409(b)(1). Under section 109(b)(2), secondary
NAAQS “shall specify a level of air quality the attainment
and maintenance of which in the judgment of the
Administrator . . . is requisite to protect the public welfare
from any known or anticipated adverse effects.” Id.
§ 7409(b)(2). The Act provides that the public welfare
protected by secondary NAAQS includes “effects on soils,
water, crops, vegetation, manmade materials, animals,
wildlife, weather, visibility, and climate, damage to and
deterioration of property, and hazards to transportation, as
well as effects on economic values and on personal comfort
and well-being.” Id. § 7602(h).

“Once EPA establishes NAAQS for a particular
pollutant, the standards become the centerpiece of a complex
statutory regime aimed at reducing the pollutant’s
 5
atmospheric concentration.” American Trucking Ass’ns v.
EPA (“ATA III”), 283 F.3d 355, 358–59 (D.C. Cir. 2002).
EPA must “complete a thorough review” of each NAAQS at
five-year intervals and “make such revisions . . . as may be
appropriate.” 42 U.S.C. § 7409(d)(1). Pursuant to section
109(d)(2), the Clean Air Scientific Advisory Committee
(“CASAC”) must periodically review NAAQS and
“recommend to [EPA] any new [NAAQS] and revisions of
existing criteria and standards as may be appropriate.” Id.
§ 7409(d)(2)(A)–(B). In proposing to issue new NAAQS or
revise existing ones, EPA must “set forth or summarize and
provide a reference to any pertinent findings,
recommendations, and comments by [CASAC]” and explain
the reasons for any “important” divergences from CASAC’s
recommendations. Id. § 7607(d)(3), (6).

These consolidated cases concern the NAAQS for ozone
(O3). Ozone is a colorless, odorless gas that is not a direct
product of human activity but instead forms when other
atmospheric pollutants react in the presence of sunlight. ATA
III, 283 F.3d at 359. EPA has identified several health effects
linked to ozone, including decreased lung function and
respiratory symptoms. Proposed National Ambient Air
Quality Standards for Ozone (“2007 Proposed Rule”), 72 Fed.
Reg. 37,818, 37,827 (July 11, 2007). EPA has also found that
ozone is associated with more serious health effects such as
increased asthma medication use, emergency department
visits, and hospital admissions. See id. at 37,827–29, 37,832.
Furthermore, EPA has determined that ozone has a broad
array of effects on trees, vegetation, and crops and can
indirectly affect other ecosystem components such as soil,
water, and wildlife. Id. at 37,883.

EPA last revised the ozone NAAQS in 1997, instituting
an “8-hour” primary standard—based on the annual fourth-
 6
highest daily maximum 8-hour average ozone
concentration—of 0.08 parts per million. National Ambient
Air Quality Standards for Ozone (“1997 Final Rule”), 62 Fed.
Reg. 38,856, 38,873 (July 18, 1997). EPA also set the
secondary NAAQS to be identical to this primary standard in
both form (measured over an 8-hour period) and level (0.08
ppm). Id. at 38,877–78. In American Trucking Associations v.
EPA, 175 F.3d 1027 (D.C. Cir. 1999), several parties
challenged these revisions, as well as the NAAQS for
particulate matter that EPA had issued at the same time. After
the Supreme Court reversed this court’s conclusion that the
Clean Air Act unconstitutionally delegated Congress’s
legislative authority, see Whitman v. American Trucking
Ass’ns, 531 U.S. 457, 473–76 (2001), we rejected all of
petitioners’ challenges to both the primary and secondary
ozone NAAQS. ATA III, 283 F.3d at 378–80.

EPA initiated the current review of the ozone NAAQS in
2000. Proceeding under a schedule adopted by consent
decree, and after receiving significant public comment on
proposed changes, EPA issued revised primary and secondary
standards on March 27, 2008. National Ambient Air Quality
Standards for Ozone (“2008 Final Rule”), 73 Fed. Reg.
16,436 (Mar. 27, 2008). In reaching its final decision, EPA
examined “the entire body of evidence relevant to examining
associations between exposure to ambient O3 and a broad
range of health endpoints.” Id. at 16,439. Of particular
relevance here, EPA emphasized new clinical studies,
including human exposure studies, showing respiratory
effects at ozone levels below 0.08 ppm. Id. at 16,449–50,
16,470–71. EPA also cited new epidemiological evidence
suggesting associations between “serious morbidity
outcomes” and ozone exposure at levels below 0.08 ppm, as
well as risk assessments estimating the effects of various
levels of ozone on the population. Id. at 16,446, 16,450–51,
 7
16,471–72. On the basis of this evidence, EPA concluded that
the existing 0.08 ppm primary standard was not requisite to
protect the public health with an adequate margin of safety.
Id. at 16,470–71.

Assessing the proper level for a revised standard, EPA
found that a level just below 0.08 ppm would be inappropriate
because “such a level would not be appreciably below the
level in controlled human exposure studies at which adverse
effects have been demonstrated.” Id. at 16,482. Although
acknowledging that CASAC had recommended a level as low
as 0.060 to 0.070 ppm, see id., EPA determined that “[a]
standard set at a level lower than 0.075 [ppm] would only
result in significant further public health protection if, in fact,
there is a continuum of health risks in areas with . . . O3
concentrations that are well below the concentrations
observed in the key controlled human exposure studies and if
the reported associations observed in epidemiological studies
are, in fact, causally related to O3 at those lower levels,” id. at
16,483. Stating that it was “not prepared to make these
assumptions,” EPA found that, with a standard set below
0.075 ppm, “the likelihood of obtaining benefits to public
health . . . decreases, while the likelihood of requiring
reductions in ambient concentrations that go beyond those
that are needed to protect public health increases.” Id.
“[J]udg[ing] that the appropriate balance to be drawn, based
on the entire body of evidence and information available in
this review, is a standard set at 0.075 [ppm],” EPA concluded
that “[a] standard set at this level provides a significant
increase in protection compared to the current standard, and is
appreciably below 0.080 ppm, the level in controlled human
exposure studies at which adverse effects have been
demonstrated.” Id.
 8
EPA also determined that the secondary standard should
be revised to be identical to the new primary standard. Id. at
16,500. Noting new evidence that had become available since
the last review, EPA found that the ozone level of the existing
secondary standard would cause significant effects on
vegetation and sensitive ecosystems. Id. at 16,496–97. EPA
acknowledged CASAC’s recommendation that a revised
secondary standard should measure ozone exposure
cumulatively over a seasonal period, rather than the 8-hour
period of the primary standard. Id. at 16,498–500. EPA
agreed with CASAC that “a cumulative, seasonal standard is
the most biologically relevant way to relate exposure to plant
growth response.” Id. at 16,500. Nonetheless, conducting a
comparison between the revised primary standard and a range
of proposed levels for a cumulative, seasonal standard, EPA
found “significant overlap between the revised 8-hour
primary standard and selected levels of the [seasonal]
standard form being considered.” Id. at 16,499. Although
recognizing that “there would be the potential for not
providing the appropriate degree of protection for vegetation
in areas with air quality distributions that result in a high
cumulative, seasonal exposure but do not result in high 8-hour
average exposures,” the agency determined that “establishing
a new secondary standard with a cumulative, seasonal form at
this time would result in uncertain benefits beyond those
afforded by the revised primary standard and therefore may
be more than necessary to provide the requisite degree of
protection.” Id. at 16,500. EPA therefore concluded that the
revised primary standard “would be sufficient to protect
public welfare from known or anticipated adverse effects,
and . . . that an alternative cumulative, seasonal standard is
[not] needed to provide this degree of protection.” Id.

Challenging the revised primary and secondary NAAQS,
various parties, including several states, the District of
 9
Columbia, New York City, and several industry,
environmental, and public health groups, filed these petitions
for review. We then granted EPA’s unopposed motion to hold
these cases in abeyance to allow the agency to review the
2008 revisions and determine whether they should be
reconsidered. In September 2011, EPA indicated that it was
withdrawing its reconsideration proceedings and would
instead be completing the reconsideration in conjunction with
the next periodic review. Several parties filed petitions for
review, challenging EPA’s withdrawal of the reconsideration
rulemaking. Finding that we lacked jurisdiction over EPA’s
non-final action, we dismissed the petitions and set a briefing
schedule for the present case.

We now confront the parties’ competing petitions for
review. One set of petitioners—comprising several states, the
District of Columbia, New York City, and a number of
environmental and public health groups—thinks the primary
and secondary NAAQS are not protective enough, while the
other set—comprising the state of Mississippi and several
industry groups—thinks they are too protective.

This opinion considers each of these claims in turn. We
reject Mississippi and the industry groups’ challenge to the
primary and secondary standards in Part II. We explain our
denial of the governmental and environmental petitions with
respect to the primary standard in Part III and our grant of
these petitions with respect to the secondary standard in Part
IV.

In considering challenges to NAAQS, “we apply the
same highly deferential standard of review that we use under
the Administrative Procedure Act.” ATA III, 283 F.3d at 362.
Accordingly, “we will set aside the Agency’s determination
only if it is ‘arbitrary, capricious, an abuse of discretion, or
 10
otherwise not in accordance with law.’ ” National
Environmental Development Ass’n’s Clean Air Project v.
EPA, 686 F.3d 803, 809–10 (D.C. Cir. 2012) (quoting 42
U.S.C. § 7607(d)(9)(A)). And “we do not look at the decision
as would a scientist, but as a reviewing court exercising our
narrowly defined duty of holding agencies to certain minimal
standards of rationality.” Id. at 810 (internal quotation marks
omitted). That said, “we perform a searching and careful
inquiry into the underlying facts.” ATA III, 283 F.3d at 362
(internal quotation marks omitted).

II.
Mississippi and the industry petitioners (collectively
“Mississippi”) challenge EPA’s threshold decision to revise
the primary NAAQS level. According to Mississippi, several
aspects of EPA’s decision were arbitrary, including its
allegedly unsupported finding that the revised NAAQS would
provide increased protection; its failure to compare the 2008
risk assessment with the 1997 risk assessment; and the
allegedly inadequate and distorted science on which the
agency relied. We reject these arguments. Mississippi also
claims the secondary NAAQS is improper because it tracks
the primary NAAQS, which Mississippi believes is unlawful,
but this argument falls as collateral damage from our rejection
of Mississippi’s challenge to the primary NAAQS.

A.
The Clean Air Act requires EPA to set primary NAAQS
that are “requisite” to protect the public health with an
adequate margin of safety. 42 U.S.C. § 7409(b)(1).
“Requisite” means the NAAQS must be “sufficient, but not
more than necessary.” Whitman, 531 U.S. at 473 (internal
quotation marks omitted). Mississippi now tells us the agency
cannot determine why further risk reduction is “requisite”
without “putting risk in the context of earlier NAAQS
 11
decisions (and other risk-based decisions).” Mississippi’s Br.
24. EPA’s failure to do so, Mississippi explains, means the
NAAQS revision is nothing more than the legally inadequate
determination that a lower level is more protective. Not so.

The force of Mississippi’s position, generated by the
Whitman Court’s malleable definition of “requisite,” assumes
only one standard at any given time can be “requisite”
because, by definition, that standard is neither higher nor
lower than necessary. Any other standard would therefore
miss the mark. But of course, this idea presupposes scientific
certainty in an area actually governed by policy-driven
approaches to uncertain science. See Lead Industries Ass’n v.
EPA, 647 F.2d 1130, 1146–47 (D.C. Cir. 1980); see also
Motor Vehicle Manufacturers Ass’n of the U.S. v. State Farm
Mutual Automobile Insurance Co., 463 U.S. 29, 57 (1983).
Mississippi’s position—though perhaps an arguable thesis—
collapses under the weight of reality.

Determining what is “requisite” to protect the “public
health” with an “adequate” margin of safety may indeed
require a contextual assessment of acceptable risk. See
Whitman, 531 U.S. at 494–95 (Breyer, J., concurring in part
and concurring in the judgment). Such is the nature of policy.
But that does not mean the initial assessment is sacrosanct and
remains the governing standard until every aspect of it is
undermined. Every time EPA reviews a NAAQS, it
(presumably) does so against contemporary policy judgments
and the existing corpus of scientific knowledge. See 42 U.S.C.
§§ 7408–09. It would therefore make no sense to give prior
NAAQS the sort of presumptive validity Mississippi insists
upon. The statutory framework requires us to ask only
whether EPA’s proposed NAAQS is “requisite”; we need not
ask why the prior NAAQS once was “requisite” but is no
longer up to the task. Following Mississippi’s approach would
 12
bind EPA to potential deficiencies in past reviews because
discrepancies between past and current judgments as easily
reflect problems in the past as in the present. We decline
Mississippi’s invitation to enter that funhouse and will defer
as long as EPA reasonably explains its actions. American
Farm Bureau Federation v. EPA, 559 F.3d 512, 521 (D.C.
Cir. 2009) (per curiam).

Mississippi argues at length that EPA should have
compared the evidence available in 2008 to the evidence
available in 1997—in particular, the clinical, epidemiological,
and toxicological studies, risk assessments, and EPA’s
protocol for sensitive populations. We need not respond point
by point; suffice to say that EPA reasonably explained how
the scientific evidence had in fact changed since the 1997
review. To name just one example, whereas in reviewing
EPA’s 1997 NAAQS-setting we emphasized “the absence of
any human clinical studies at ozone concentrations below
0.08,” ATA III, 283 F.3d at 379, EPA here explained that “two
new controlled human-exposure studies . . . are now available
that examine respiratory effects associated with prolonged O3
exposures at levels at and below 0.080 ppm.” 2008 Final
Rule, 73 Fed. Reg. at 16,454. 1 But to frame it more broadly,

1
Because Mississippi independently challenges EPA’s failure
to compare its 2008 and 1997 risk assessments, however, we also
acknowledge the reasonableness of EPA’s explanation for not
doing so—namely, that a comparison would be “factually
inappropriate,” would not account for the fact that “with similar
risks, increased certainty in the risks” would engender greater
concern, and would obscure the qualitativeness of EPA’s approach.
2008 Final Rule, 73 Fed. Reg. at 16,466. First, the 2008 risk
assessment analyzed a number of health effects not included in the
1997 risk assessment, so the ultimate value of comparing the two
assessments would be limited. Second, logic rejects comparisons of
 13
we note, first, that the NAAQS review process includes
EPA’s public health policy judgments as well as its analysis
of scientifically certain fact, and, second, that as the contours
and texture of scientific knowledge change, the
epistemological posture of EPA’s NAAQS review necessarily
changes as well; additional certainty about what was merely a
thesis might very well support a determination that the line
marked by the term “requisite” has shifted. In short,
Mississippi seeks to eliminate any adumbration of the
inevitable scientific uncertainties that shadow and shape
EPA’s statutory mandate to take a preventative approach. See,
e.g., ATA III, 283 F.3d at 378.

B.
Mississippi looses the rest of its arrows at the evidence
on which EPA relied, although in doing so, Mississippi again
showcases its apparent preference for exuberance over
precision. Ultimately, Mississippi’s arguments that the 2008
science added nothing new to the 1997 NAAQS conversation
and that EPA misrepresented the science on which it relied

apples and oranges, which is how we would describe two data
analyses subject to different geographic and demographic
parameters. See, e.g., 2007 Proposed Rule, 72 Fed. Reg. at 37,851–
52 (explaining that 2008 exposure analysis, an input to part of the
risk assessment, relied on a model different from the one used in
1997). And finally, even if EPA should have compared the two risk
assessments where they overlapped, EPA’s failure to do so does not
necessarily render EPA’s ultimate NAAQS decision improper. See
ATA III, 283 F.3d at 369–70; see also Lead Industries Ass’n, 647
F.2d at 1162. The risk assessment turned on more than just those
risks amenable to comparison with the 1997 risk assessment, and in
setting the NAAQS, EPA relied on much more than just the risk
assessment. See, e.g., 2008 Final Rule, 73 Fed. Reg. at 16,467,
16,476, 16,479.
 14
are largely dependent on the conceptual error that EPA is
somehow bound by the 1997 NAAQS and on the legal error
that it is our job to “weigh the evidence anew.” American
Farm Bureau, 559 F.3d at 533 (internal quotation marks
omitted). Nonetheless, we address each argument in turn,
construing Mississippi’s arguments to articulate a weightier
challenge: that the available evidence did not support EPA’s
threshold decision to revise the NAAQS. We again disagree.

1.
The 1997 standard was “set at a level of 0.08 ppm, which
is equivalent to 0.084 ppm using the standard rounding
conventions.” 2008 Final Rule, 73 Fed. Reg. at 16,437. By
framing the issue in terms of EPA’s decision to lower the
NAAQS level below 0.080 ppm, Mississippi fails to capture
the full significance of the 2008 NAAQS revision. 2 Indeed,
Mississippi’s imprecision undermines its case: by conceding
that health effects are linked to ozone levels of 0.080 ppm,
Mississippi rebuts its claim that science in 2008 did not
support a NAAQS set at an effective level lower than 0.084
ppm.

In any event, after reviewing the record, we think it quite
clear EPA’s rejection of the 1997 NAAQS was proper. EPA
relied on a broad array of scientific studies, quantified
models, and input from CASAC, EPA staff, and commenters,
and it considered not only what was known, but also what was
not known. See, e.g., 2008 Final Rule, 73 Fed. Reg. at
16,438–40. It then evaluated the evidence as a whole through
2
For that matter, EPA commits the same error, referring to
“0.080” when it should refer to “0.08,” compare 2008 Final Rule,
73 Fed. Reg. at 16,444, with 1997 Final Rule, 62 Fed. Reg. at
38,859, but Mississippi does not note it, so we have no reason to
think this reflects anything other than sloppy editing.
 15
an “integrative synthesis,” what it called a “weight of
evidence approach.” Id. at 16,439, 16,479. And appropriately
so: one type of study might be useful for interpreting
ambivalent results from another type, see Ethyl Corp. v. EPA,
541 F.2d 1, 26 (D.C. Cir. 1976) (en banc), and though a new
study does little besides confirm or quantify a previous
finding, such incremental (and arguably duplicative) studies
are valuable precisely because they confirm or quantify
previous findings or otherwise decrease uncertainty. See, e.g.,
2008 Final Rule, 73 Fed. Reg. at 16,450 (noting that post-
1997 evidence “increased the Administrator’s confidence”
that particular health endpoints are causally related to ozone
exposure). EPA made this point when it explained that
controlled human exposure studies provide “the most directly
applicable” evidence (and engender “the highest level of
confidence”) about the causal relationship between ozone
exposure and health effects; that epidemiological studies
provide evidence both about health effects from exposure to
ambient air and about the effect of ozone exposure on “more
serious” health effects like hospital admissions and mortality;
and that animal toxicology studies generally support the
biological plausibility of effects noted in clinical and
epidemiological studies. 2007 Proposed Rule, 72 Fed. Reg. at
37,823, 37,825; see 2008 Final Rule, 73 Fed. Reg. at 16,440
(incorporating discussion of scientific evidence in proposed
rule). Given that the record includes, among other things,
numerous epidemiological studies linking health effects to
exposure to ozone levels below 0.08 ppm and clinical human
exposure studies finding a causal relationship between health
effects and exposure to ozone levels at and below 0.08 ppm,
we will not second-guess EPA’s interpretations of, or the
conclusions it drew from, this evidence.

Reasonable people might disagree with EPA’s
interpretations of the scientific evidence, but any such
 16
disagreements must come from those who are qualified to
evaluate the science, not us. We are satisfied that EPA’s
interpretations are permissible, and that is enough. Indeed,
CASAC unanimously concluded that “[t]here is no scientific
justification for retaining the current primary 8-hr NAAQS of
0.08 parts per million,” that the primary NAAQS “needs to be
substantially reduced to protect human health,” and that the
primary NAAQS should be set at a level somewhere between
0.060 and 0.070 ppm. Letter from Dr. Rogene Henderson,
CASAC Chair, to Stephen L. Johnson, EPA Administrator
(“Oct. 2006 CASAC Letter”), at 1–2 (Oct. 24, 2006), EPA-
CASAC-07-001. If, as we have explained, EPA may give
“significant weight” to propositions about the appropriate
NAAQS level implicitly accepted by otherwise-disagreeing
CASAC members, see ATA III, 283 F.3d at 378–79, surely it
may rely on an explicit recommendation by the unanimous
CASAC panel.

And given the reasonableness of EPA’s interpretation of
the science, its determination that the 1997 NAAQS was
insufficiently protective of public health follows as a matter
of course. EPA concluded sensitive populations like
asthmatics are affected by ozone in a more severe way and at
lower levels than are healthy individuals and that ozone-
related health effects might be adverse for sensitive
individuals though comparable effects would not be
considered adverse for healthy individuals—conclusions
Mississippi does not challenge. See 2008 Final Rule, 73 Fed.
Reg. at 16,454–55, 16,466. EPA could properly decide that a
NAAQS set at the level of 0.08 ppm does not protect the
public health with an adequate margin of safety when healthy
individuals experience adverse health effects from exposure
to ozone at and below that level. See, e.g., American Farm
Bureau, 559 F.3d at 525–26. “That petitioners . . . find a basis
to disagree” with EPA is “hardly surprising.” Ethyl Corp., 541
 17
F.2d at 26. But that does not make EPA’s decision to revise
the NAAQS arbitrary.

2.
Mississippi also contends that section 108(a)(2) of the
Clean Air Act, 42 U.S.C. § 7408(a)(2), requires EPA “to
consider and rely upon all scientific information that is
capable of being put to use and serviceable for [identifying
the effect that a given pollutant has on public health and
welfare] and that is free from error (accurate).” Mississippi’s
Br. 47. This requirement, Mississippi explains, incorporates
information standards under the Information Quality Act
(“IQA”), Pub. L. No. 106-554, sec. 1(a)(3), § 515, 114 Stat.
2763, 2763A-153 to 154 (2000) (H.R. 5658) (codified at 44
U.S.C. § 3516 note). According to Mississippi (as we
understand its argument), EPA violated both the Clean Air
Act and the IQA by inaccurately characterizing some studies
and by relying on other, flawed studies. These arguments are
difficult to parse because they fill two roles in Mississippi’s
shadow play: reinforcing the implied point that the available
evidence did not support EPA’s decision to revise the
NAAQS level and unmasking a violation of the Clean Air
Act’s procedural requirements. Having already discussed the
reasonableness of EPA’s threshold decision to revise the
NAAQS, we now consider only Mississippi’s suggestion that
EPA violated the Clean Air Act’s and the IQA’s procedural
standards and that these violations independently render
EPA’s NAAQS determination unlawful.

The Clean Air Act implicitly divides the NAAQS review
process into three stages. First, members of the scientific
community publish studies, which may be more or less
flawed. Second, EPA must issue and periodically revise air
quality criteria that “accurately reflect the latest scientific
knowledge useful in indicating the kind and extent of all
 18
identifiable effects on public health or welfare which may be
expected from the presence of such pollutant in the ambient
air, in varying quantities.” 42 U.S.C. §§ 7408(a)(2), 7409(d).
Finally, EPA must “base[]” its NAAQS determinations on the
criteria. 42 U.S.C. § 7409(b)(1). From start to finish, this
system is vulnerable to error. In particular, even if the
foundational scientific studies are not flawed in any material
way, transmission errors may nevertheless occur when EPA
drafts the criteria or when it subsequently decides what
NAAQS to set. Mississippi’s point appears to be that the
Clean Air Act and the IQA impose safeguards to ensure
accuracy throughout this entire process. While that may be a
fair characterization, it overstates the practical effect of the
statutory schemes.

First, though the Clean Air Act requires the air quality
criteria to “accurately reflect” the scientific evidence, that
requirement says nothing about the accuracy of the science
itself or the precision of the relationship between the criteria
and EPA’s NAAQS decision. The criteria, which are neither
“standards” nor “guidelines,” simply “provide the scientific
basis for promulgation of air quality standards.” Lead
Industries Ass’n, 647 F.2d at 1136–37. We do not reweigh the
evidence or second-guess technical judgments but are limited
to determining whether EPA made a rational judgment. See
American Petroleum Institute v. Costle, 665 F.2d 1176, 1185
(D.C. Cir. 1981). Nor do we look through the microscope to
scrutinize EPA’s use of the criteria: there are limits to EPA’s
discretion in using the criteria, see Michigan v. EPA, 213 F.3d
663, 696 (D.C. Cir. 2000) (per curiam), but EPA’s translation
of the criteria into a NAAQS decision is not frictionless, and
ignoring this fact would squeeze considerations of policy and
the role of CASAC out of the equation. See 42 U.S.C.
§ 7607(d)(3); see also Catawba County, North Carolina v.
EPA, 571 F.3d 20, 35 (D.C. Cir. 2009) (per curiam).
 19

Second, Mississippi fails to show the IQA is an
independent measure of EPA’s NAAQS decision. The IQA
requires the Director of the Office of Management and
Budget to provide “policy and procedural guidance” to
“ensur[e] and maximiz[e] the quality, objectivity, utility, and
integrity of information . . . disseminated by Federal
agencies.” 44 U.S.C. § 3516 note. OMB, in turn, issued
flexible, “generic” guidelines that it recognized “cannot be
implemented in the same way by each agency.” Guidelines
for Ensuring and Maximizing the Quality, Objectivity, Utility,
and Integrity of Information Disseminated by Federal
Agencies; Republication, 67 Fed. Reg. 8,452, 8,452–53 (Feb.
22, 2002). EPA’s implementing guidelines, meanwhile,
purport to provide only “non-binding policy and procedural
guidance.” EPA, GUIDELINES FOR ENSURING AND
MAXIMIZING THE QUALITY, OBJECTIVITY, UTILITY, AND
INTEGRITY OF INFORMATION DISSEMINATED BY THE
ENVIRONMENTAL PROTECTION AGENCY 4 (2002). Mississippi
points to nothing indicating that any part of this scheme
committed EPA to having done things differently. See
American Petroleum Institute v. EPA, 684 F.3d 1342, 1348
(D.C. Cir. 2012); see also Salt Institute v. Leavitt, 440 F.3d
156, 159 (4th Cir. 2006).

Measuring Mississippi’s challenge to EPA’s use of the
scientific evidence against the agency’s legal obligations, we
see nothing to suggest EPA acted improperly, particularly
given our approval of its ultimate decision to revise the
NAAQS. To start, Mississippi’s challenge to EPA’s use of the
Adams studies—a set of controlled human exposure studies
that played a relatively significant role in the NAAQS review
process—is nothing more than a claim that EPA did wrong by
disagreeing with Adams’s interpretation of his data.
(Although Adams concluded his 2006 study did not show
 20
statistically significant lung function decrements at the 0.06
ppm ozone exposure level, EPA explained that it believed a
different statistical model would more directly address the
precise question with which it was concerned—namely, the
effects of prolonged ozone exposure versus exposure to
filtered air—and that application of this model yielded
statistically significant results at the 0.06 ppm level.) Yet
nothing in the Clean Air Act or the IQA prohibits EPA from
independently analyzing the science—for example, by asking
“different questions” from those asked by the study’s author,
2008 Final Rule, 73 Fed. Reg. at 16,455—and the only
objections Mississippi offers to EPA’s independent analysis
are either conclusory or require us to weigh in on what is
apparently legitimate scientific debate. See id. (noting
approval by members of CASAC of the statistical approach
used in the reanalysis). 3

Mississippi’s challenge to EPA’s use of the
epidemiological evidence fares no better. Though it claims
EPA improperly relied on studies using ambient ozone data as
a proxy for personal exposure, Mississippi neither challenges
EPA’s explanation that very few epidemiological studies
directly measuring personal exposure exist in the literature
nor acknowledges EPA’s recognition that ambient
measurements do not necessarily represent personal exposure
levels and must therefore be used with caution. We have no
problem with EPA’s reliance on actual, rather than

3
To the extent Mississippi’s complaint centers around EPA’s
failure to peer review its reanalysis, we note that EPA’s IQA
guidelines expressly disclaim a categorical peer-review policy, so
even assuming Mississippi is right that the reanalysis was not peer
reviewed, Mississippi’s failure to explain why the alleged lack of
peer review was improper is fatal. See American Petroleum
Institute, 684 F.3d at 1348–49.
 21
nonexistent, evidence, and in any event, Mississippi does not
challenge EPA’s interpretation of the measurement
disparity—that if the disparity biases the epidemiological
evidence, it does so by underestimating ozone’s health
effects. See 2007 Proposed Rule, 72 Fed. Reg. at 37,839.
Finally, Mississippi’s belief that EPA ignored contradictory
evidence is an example of its own confirmation bias.
Mississippi insists EPA failed to account for studies
suggesting that findings of ozone-related health effects may
be confounded by the presence of other pollutants, but this
challenge boils down to a claim about two epidemiological
studies. EPA mentions only one of these studies—and only
once—in the final rule, and in doing so, it also cites two other
studies (which Mississippi does not challenge) for the same
proposition. See 2008 Final Rule, 73 Fed. Reg. at 16,446.
Even granting the substance of Mississippi’s assertions
(which we do not), it is hard to imagine how eliminating both
studies from EPA’s NAAQS calculation would have altered
EPA’s ultimate decision. 4

We repeat: it is not our job to referee battles among
experts; ours is only to evaluate the rationality of EPA’s
decision, and as we have explained, the agency did its part.
And because we reject Mississippi’s challenge to the primary
NAAQS, we must also reject its challenge to the secondary
NAAQS. Mississippi’s petition for review is therefore denied.

III.
As discussed above, EPA’s review of the available
scientific evidence led it to adopt a primary ozone NAAQS of

4
EPA also cited the two studies a total of five times in the
proposed rule, but we think this immaterial in light of Mississippi’s
failure to explain their importance to EPA’s final decision.
 22
0.075 ppm. While Mississippi criticized EPA’s decision to
reduce this standard from its prior level of 0.08 ppm, multiple
state and local governments, environmental advocacy non-
profits, and public health non-profits contend that EPA did
not go far enough. Thus, EPA finds itself in a situation
reminiscent of Goldilocks and the Three Bears. On one side,
Mississippi argued that EPA is too stringent with its ozone
NAAQS; on the other side, the governmental and
environmental petitioners argue that the NAAQS is too lax.
But unlike Goldilocks, this court cannot demand that EPA get
things “just right.” Rather, for EPA’s decision to survive these
challenges, it need do no more than meet the statutory
standards found in the Clean Air Act. “That the evidence in
the record may also support other conclusions, even those that
are inconsistent with [EPA’s], does not prevent us from
concluding that [its] decisions were rational . . . .” Lead
Industries Ass’n, 647 F.2d at 1160 (footnote and citations
omitted).

The Act requires us to overturn any EPA action that is
arbitrary, capricious, an abuse of discretion, or contrary to
law. 42 U.S.C. § 7607(d)(9)(A). The governmental and
environmental petitioners argue that EPA’s judgment—that a
primary ozone NAAQS of 0.075 ppm is “requisite to protect
the public health,” 42 U.S.C. § 7409(b)(1)—was arbitrary and
capricious because EPA failed to rationally consider scientific
evidence demonstrating adverse health effects at ozone levels
below 0.075 ppm. They also argue that EPA acted contrary to
law because it failed to calculate an adequate margin of
safety, as required by section 109(b)(1) of the Act, 42 U.S.C.
§ 7409(b)(1). Finally, they argue that EPA violated its
statutory duty to explain and defend its decision to depart
from CASAC’s recommendations. See id. § 7607(d)(3), (6).
We address each of these arguments in turn.
 23
A.
It is true that “[a]n agency’s failure adequately to
consider a relevant and significant aspect of a problem may
render its rulemaking arbitrary and capricious.” American
Farm Bureau, 559 F.3d at 520. But the corollary to EPA’s
obligation to “weigh the entire record,” Achernar
Broadcasting Co. v. FCC, 62 F.3d 1441, 1446 (D.C. Cir.
1995), is that no single piece of evidence is dispositive. See
American Farm Bureau, 559 F.3d at 525; see also ATA III,
283 F.3d at 379. Moreover, “we do not determine the
convincing force of evidence, nor the conclusion it should
support, but only whether the conclusion reached by EPA is
supported by substantial evidence when considered on the
record as a whole.” Coalition for Responsible Regulation, Inc.
v. EPA, 684 F.3d 102, 122 (D.C. Cir. 2012) (per curiam).

Provided EPA meets its obligation “to explain and
expose every step of its reasoning,” American Lung Ass’n v.
EPA, 134 F.3d 388, 392 (D.C. Cir. 1998), the governmental
and environmental petitioners have a heavy burden to show
that the totality of the evidence required EPA to decide
differently than it did. Lead Industries Ass’n, 647 F.2d at
1160. This approach to giant administrative records is
consistent with the deference principles discussed above. See
Part II, supra at 14. Our role is circumscribed. We are merely
to “determin[e] if [EPA] made a rational judgment,” not to
“weigh the evidence anew and make technical judgments.”
Costle, 665 F.2d at 1185.

The governmental and environmental petitioners argue
that EPA failed to grapple with three major types of evidence
that they claim favor a lower primary ozone NAAQS:
controlled human exposure studies, epidemiological studies,
and human exposure and health risk assessments. The record
reveals, and petitioners do not dispute, that EPA considered
 24
the entire body of scientific evidence available to it,
discussing each type of evidence at each stage of its analysis.
See, e.g., 2007 Proposed Rule, 72 Fed. Reg. at 37,864–68;
2008 Final Rule, 73 Fed. Reg. at 16,452–70; see also id. at
16,439 (describing the range of evidence considered in the
years-long review process). The petitioners argue instead that
EPA’s conclusion that a level of 0.075 ppm is “requisite” to
protect public health cannot be rationally drawn from this
evidence. We disagree. EPA’s treatment of the evidence
satisfies our deferential standard of review.

Petitioners argue that the controlled human exposure
studies—in particular, the Adams studies—support a more
protective primary NAAQS because they demonstrate adverse
effects at the 0.060 ppm level. The Adams studies, published
in 2002 and 2006, analyzed the results of laboratory
experiments that directly measured the effects of ozone on
humans’ respiratory health by exposing thirty subjects to
ozone in a controlled environment. Adams tested his subjects
at various ozone concentrations, including 0.08 ppm and 0.06
ppm, but at no levels in between. He found that a small
number of subjects exposed to ozone at 0.06 ppm experienced
lung function decrements of at least ten percent—a level EPA
considers to be harmful (or “adverse”) to asthmatics. See 2008
Final Rule, 73 Fed. Reg. at 16,454–55. Petitioners argue that
the 0.06 ppm Adams results were “unrebutted ‘substantial
evidence’ ” favoring a lower standard, and that EPA’s
decision to set the standard as high as 0.075 ppm “ ‘is not
supported by substantial evidence.’ ” Environmental
Petitioners’ Br. 19 (quoting City of Naples Airport Authority
v. FAA, 409 F.3d 431, 436 (D.C. Cir. 2005)). The crux of the
dispute is whether EPA rationally treated this evidence of
adverse effects as not dispositive.
 25
EPA relied on the Adams studies and other clinical
studies to justify its decision to lower the primary ozone
NAAQS from the 0.08 ppm level, concluding that they
“provide[d] the most certain evidence of adverse health
effects” at 0.080 ppm available. 2008 Final Rule, 73 Fed.
Reg. at 16,478. EPA also conducted a reanalysis of the
Adams (2006) study that found “small group mean
decrements in lung function responses to be statistically
significant at the 0.060 ppm exposure level.” Id. at 16,454.
But EPA further concluded that the data at the 0.060 ppm
level was too limited to support a reduction in the NAAQS to
that level.

Each Adams study involved only thirty subjects, of which
six at most experienced lung function decrements of ten
percent or more at exposure levels below 0.080 ppm. 5 For this
reason, the CASAC scientists had mixed views about the
Adams studies. For instance, one scientific advisor stated that
the number of data points in the Adams studies was “pitiful,”
and that the limited nature of the data was “astounding.”
Letter from Dr. Rogene Henderson, CASAC Chair, to

5
The record includes conflicting accounts of the number of
participants experiencing lung function decrements of ten percent
or larger at 0.06 ppm. All accounts indicate that it was a small
number, and never more than six. The precise number appears to
depend on one’s method of measuring lung function decrements.
Compare OFFICE OF AIR QUALITY PLANNING AND STANDARDS,
EPA, REVIEW OF THE NATIONAL AMBIENT AIR QUALITY
STANDARDS FOR OZONE: POLICY ASSESSMENT OF SCIENTIFIC AND
TECHNICAL INFORMATION (“STAFF PAPER”), § 3.3.1.1.1 (2007),
EPA-452/R-07-007, with Letter from Dr. Rogene Henderson,
CASAC Chair, to Stephen L. Johnson, EPA Administrator (“Mar.
2007 CASAC Letter”), at C-31–32 (Mar. 26, 2007), EPA-CASAC-
07-002.
 26
Stephen L. Johnson, EPA Administrator (“Mar. 2007 CASAC
Letter”), at C-31–32 (Mar. 26, 2007), EPA-CASAC-07-002.
Another cautioned that the responses Adams recorded at the
0.06 ppm level might merely reflect “normal variations” in
human lung function rather than “real ozone responses.” Oct.
2006 CASAC Letter, at D-14. In other words, other factors
apart from a change in ozone levels—for example, participant
fatigue or diminished effort—might explain the “decrements”
that Adams observed. Adams himself was critical of those
who drew strong conclusions from his results at the 0.06 ppm
level because he determined that the average responses were
not statistically significant. See William C. Adams, Comment
on EPA Memorandum: The Effects of Ozone on Lung
Function at 0.06 ppm in Healthy Adults (Oct. 9, 2007), EPA-
HQ-OAR-2005-0172-4783. Ultimately, although EPA
disagreed with Adams regarding the statistical significance of
some results, it found that the study’s small sample size could
not “appropriately be generalized to the U.S. population.”
2008 Final Rule, 73 Fed. Reg. at 16,454, 16,478.

Thus, while the 0.08 ppm results were robust, EPA
rationally treated the 0.06 ppm results as inconclusive.
Perhaps more studies like the Adams studies will yet reveal
that the 0.060 ppm level produces significant adverse
decrements that simply cannot be attributed to normal
variation in lung function. But at the time of EPA’s
rulemaking, it was rational to treat the 0.06 ppm results with
skepticism. The Adams results at 0.06 ppm indicate some
degree of risk that some number of individuals might continue
to experience health effects at and below 0.075 ppm, but we
have previously acknowledged the impossibility of
eliminating all risk of health effects from “non-threshold”
pollutants like ozone. See ATA III, 283 F.3d at 360 (“The lack
of a threshold concentration below which these pollutants are
known to be harmless makes the task of setting primary
 27
NAAQS difficult, as EPA must select standard levels that
reduce risks sufficiently to protect public health even while
recognizing that a zero-risk standard is not possible.” (internal
quotation marks and original alterations omitted)).

Petitioners counter that EPA has relied on even
statistically nonsignificant results in the past when setting the
primary ozone NAAQS, so the limitations of the Adams
studies provide no basis for dismissing the evidence of
adverse effects at that level. Environmental Petitioners’ Br.
20–21. Be that as it may, the question for this court is not
what EPA has done in the past, or even what levels it
rationally could have settled on, but only whether it has
provided a rational explanation of how it treated the evidence
before it. See ATA III, 283 F.3d at 374 (“[W]e review [EPA’s]
scientific judgments . . . not as the chemist, biologist, or
statistician that we are qualified neither by training nor
experience to be, but as a reviewing court exercising our
narrowly defined duty of holding agencies to certain minimal
standards of rationality.” (quoting Troy Corp. v. Browner, 120
F.3d 277, 283 (D.C. Cir. 1997)). Statistical quality affords a
perfectly rational basis for assigning different weights to
different pieces of scientific data when evaluating the totality
of the evidence. While EPA is certainly permitted to look to
statistically uncertain results, it is by no means required to
rely on them. Its failure to do so in this case did not render its
decision irrational.

The governmental and environmental petitioners next
argue that EPA gave short shrift to the epidemiological
studies. By using statistical techniques to analyze vast bodies
of health and environmental data across large populations,
epidemiological studies allow scientists to draw inferences
about the harms of ozone without carefully calibrated
laboratory experiments. EPA relied on over 250 such studies
 28
during its 2008 rulemaking. 2008 Final Rule, 73 Fed. Reg. at
16,455, 16,479. Petitioners point out that some studies found
significant correlations between ozone concentration and
adverse health outcomes at levels well below 0.075 ppm. See
Comments of the American Lung Association, Environmental
Defense, Sierra Club on the U.S. Environmental Protection
Agency’s Proposed Revisions to the NAAQS for Ozone (Oct.
9, 2007), EPA-HQ-OAR-2005-0172-4261; Comments of
American Thoracic Society, et al. (Oct. 9, 2007), EPA-HQ-
OAR-2005-0172-4305. The studies were relatively consistent,
and the results—as EPA admits—may help establish a causal
relationship between the presence of ozone and the
occurrence of adverse health effects. 2008 Final Rule, 73 Fed.
Reg. at 16,450.

As with the Adams studies, EPA relied on the
epidemiological studies to conclude that the existing standard
of 0.08 ppm was too high. EPA noted that many
epidemiological studies reported “statistically significant
associations that generally extend down to ambient O3
concentrations that are below the level of the current
standard” and considered these studies as part of the body of
“new evidence demonstrating that exposures to O3 at levels
below the level of the current standard are associated with a
broad array of adverse health effects.” Id. at 16,471. EPA also
explained, however, that “the epidemiological studies are not
themselves direct evidence of a causal link between exposure
to O3 and the occurrence of [health] effects,” id. at 16,479,
and that evidence of this causal relationship “becomes
increasingly uncertain at lower levels of exposure.” Id. at
16,478. EPA explained this uncertainty by reference to
intrinsic indicators of reliability and extrinsic sources of
corroboration, both of which provide substantial evidence for
EPA’s decision. For example, at much lower levels of ozone
exposure, EPA questioned whether it could attribute the
 29
epidemiological effects to ozone alone “rather than to the
broader mix of air pollutants present in the ambient air.” Id. at
16,456; see also EPA, Responses to Significant Comments on
the 2007 Proposed Rule, at 29 (Mar. 2008), EPA-HQ-OAR-
2005-0172-7185. Additionally, EPA relied on controlled
studies like the Adams studies to lend “biological plausibility”
to the inferences of causation drawn from epidemiological
studies. According to EPA, while “[t]he biological plausibility
of the epidemiological associations is generally supported by
controlled human exposure and toxicological evidence of
respiratory morbidity effects for levels at and below 0.080
ppm,” that “biological plausibility becomes increasingly
uncertain at much lower levels.” 2008 Final Rule, 73 Fed.
Reg. at 16,456. EPA’s discussion of the limitations of the
epidemiological studies at lower levels of ozone exposure
satisfies the “minimal standards of rationality” to which we
hold the agency. See National Environmental Development
Ass’n’s Clean Air Project, 686 F.3d at 810 (internal quotation
marks omitted).

Petitioners also challenge EPA’s interpretation of its own
risk and exposure assessments. EPA did not rely heavily on
them, though petitioners think it should have. These
assessments model real-world interactions between a host of
variables in order to predict health outcomes based on
available data. 2008 Final Rule, 73 Fed. Reg. at 16,441. As
such, they adhere to the inviolable law of data analysis,
“garbage in; garbage out.” That is, as CASAC cautioned EPA,
the risk and exposure assessments are only as reputable as the
inputs upon which they rely to produce their predictions. Oct.
2006 CASAC Letter, at 12; see also Letter from Dr. Rogene
Henderson, CASAC Chair, to Stephen L. Johnson, EPA
Administrator, at D-39 (Feb. 10, 2006), EPA-CASAC-06-003
(discussing similar weaknesses in risk assessment for the
secondary NAAQS). In this case, the inputs were the very
 30
data whose reliability EPA questioned at lower levels.
Recognizing their limitations, we have previously approved
EPA’s cautious treatment of risk and exposure assessments
when EPA “consider[s] all aspects of the problem” and
“catalogue[s] its concerns.” See American Farm Bureau, 559
F.3d at 527. We do the same now.

Having reasonably explained the limitations it believed
existed in each of these bodies of scientific evidence, EPA
concluded that the standard “must be set at a level appreciably
below 0.080 ppm, the level at which there is considerable
evidence of effects in healthy people.” 2008 Final Rule, 73
Fed. Reg. at 16,480; see also id. at 16,483 (“0.080 ppm [is]
the level in controlled human exposure studies at which
adverse effects have been demonstrated.”). EPA concluded
that a standard set at 0.075 ppm “would be requisite to protect
public health with an adequate margin of safety, including the
health of sensitive subpopulations.” Id. at 16,483. EPA
explained that a standard lower than 0.075 ppm was not
required because it “would only result in significant further
public health protection if, in fact, there is a continuum of
health risks in areas with 8-hour average O3 concentrations
that are well below the concentrations observed in the key
controlled human exposure studies and if the reported
associations observed in epidemiological studies are, in fact,
causally related to O3 at those lower levels.” Id. Based on the
uncertainties EPA had identified “in interpreting the evidence
from available controlled human exposure and
epidemiological studies at very low levels,” EPA was “not
prepared to make these assumptions.” Id. Finding that “the
likelihood of obtaining benefits to public health with a
standard set below 0.075 ppm O3 decreases, while the
likelihood of requiring reductions in ambient concentrations
that go beyond those that are needed to protect public health
increases,” EPA judged that “the appropriate balance to be
 31
drawn” was a standard set at 0.075 ppm. Id. We see nothing
arbitrary and capricious about EPA’s balancing of these
considerations.

B.
The governmental and environmental petitioners next
argue that, even if the scientific evidence of adverse effects at
ozone levels below 0.075 ppm remained uncertain, the
overwhelming evidence of adverse effects at 0.080 ppm
required a primary NAAQS lower than 0.075 ppm to ensure
an adequate margin of safety. EPA is required to “allow[] an
adequate margin of safety” in setting a primary NAAQS that
is “requisite to protect the public health.” 42 U.S.C.
§ 7409(b)(1). By requiring an “adequate margin of safety,”
Congress was directing EPA to build a buffer to protect
against uncertain and unknown dangers to human health.
Lead Industries Ass’n, 647 F.2d at 1154; see also ATA III, 283
F.3d at 368. Our case law has left EPA with a wide berth
when it comes to deciding how best to account for an
adequate margin of safety. In Lead Industries Association, we
held that the choice of how to set a margin of safety is “a
policy choice of the type that Congress specifically left to the
Administrator’s judgment.” 647 F.2d at 1162. And in
American Trucking Associations, we clarified that EPA need
not “identify[] a ‘safe level’ and then apply[] an additional
margin of safety”; instead, it may “take into account margin
of safety considerations throughout the process as long as
such considerations are fully explained and supported by the
record.” ATA III, 283 F.3d at 368 (internal quotation marks
omitted).

In light of this deferential standard, we have only rarely
found that the agency failed to build in a margin of safety.
See, e.g., American Farm Bureau, 559 F.3d at 525–26
(granting the petition for review in part because EPA failed to
 32
account for a margin of safety). When we have, it has not
been on the basis of our own untutored judgment about how
large a margin is necessary, but rather for egregious
procedural errors, such as EPA’s failure to consider sensitive
sub-populations, like asthmatics, children, or the elderly. See
id. In this case, no such problem presents itself; EPA regularly
and consistently considered the effects of its rules on these
sensitive groups. See, e.g., 2008 Final Rule, 73 Fed. Reg. at
16,476. EPA acknowledged that some of these subpopulations
are more likely to experience adverse effects at all levels of
exposure, requiring it to select a primary NAAQS level below
the level at which adverse effects occur “with reasonable
scientific certainty.” See id. at 16,437 (explaining the purpose
of the margin of safety); id. at 16,449 (describing CASAC’s
conclusion that existing studies do not adequately cover
sensitive subpopulations); id. at 16,452 (adopting that
conclusion in part). As a result, EPA set the standard
“appreciably below” 0.080 ppm, the lowest level at which
EPA expressed confidence that ozone causes adverse health
effects in healthy individuals. Id. at 16,480. Petitioners have
given us no reason to doubt EPA’s characterization of the
0.075 ppm level as “appreciably below” 0.080 ppm. EPA
complied with Congress’s command in section 109(b)(1) to
build in a margin of safety, and its judgment that this margin
is adequate was not arbitrary or capricious.

C.
The governmental and environmental petitioners next
argue that EPA failed to uphold its duty under the Act to
provide “an explanation of the reasons” for departing from
CASAC’s recommendations. 42 U.S.C. § 7607(d)(3); see also
id. § 7607(d)(6)(A). Congress created CASAC in the 1977
Clean Air Act Amendments and tasked it with providing
scientific advice to aid EPA in setting NAAQS. See id.
§ 7409(d)(2). Expressing its “desire for continued
 33
independent scientific review of the Environmental Protection
Agency’s exercise of judgment,” H. Rep. No. 95-294, at 182
(1977), Congress directed CASAC to complete a review of
the air quality criteria and primary and secondary NAAQS
every five years and to “recommend to the Administrator any
new national ambient air quality standards and revisions of
existing criteria and standards as may be appropriate,” 42
U.S.C. § 7409(d)(2)(B).

When Congress created CASAC, the promulgation of
NAAQS was in its infancy. In describing the role it
envisioned for CASAC, Congress emphasized the valuable
role that advisory committees and expert groups had played in
reviewing the first criteria documents and air quality
standards issued in the late 1960s and early 1970s, explaining
that “[f]or nearly 10 years the scientific basis for setting
ambient air quality standards has been reviewed, evaluated,
subjected to outside criticism, and reevaluated.” H. Rep. No.
95-294, at 179–81. CASAC was intended to replicate this role
by “provid[ing] an independent source of review and advice
to the Administrator and to the Congress.” Id. at 182. Thus,
Congress explained that it established CASAC “[b]ecause of
the admitted need for greater research, the importance of the
national ambient air quality standards, the continuing
controversy over the standards, and the committee’s desire for
continued independent scientific review of the Environmental
Protection Agency’s exercise of judgment.” Id.

Congress expected that CASAC’s central role would be
one of scientific analysis, explaining that CASAC’s “main
function” was “to assess the health and environmental effects
of ambient air pollution.” Id. at 183. CASAC would “provide
an outside mechanism for evaluating whether any pollutant
may reasonably be anticipated to endanger public health or
environment, for evaluating the scientific and medical data
 34
which might bear on this question, and for reviewing gaps in
the available data and recommending additional needs for
research.” Id. at 182. Given these functions, Congress
expected that CASAC members would “be selected on the
basis of their special expertise” in fields such as
“environmental toxicology, epidemiology and/or clinical
medicine.” Id. at 183.

Congress also required EPA to take CASAC’s expert
scientific recommendations into account in promulgating
NAAQS. Although EPA is not bound by CASAC’s
recommendations, it must fully explain its reasons for any
departure from them. Specifically, section 307(d)(3) of the
Act mandates that when EPA proposes to issue new NAAQS
or revise existing NAAQS, the proposed rule must include a
“statement of its basis and purpose” that “set[s] forth or
summarize[s] and provide[s] a reference to any pertinent
findings, recommendations, and comments by [CASAC].” 42
U.S.C. § 7607(d)(3). If EPA’s “proposal differs in any
important respect from any of [CASAC’s] recommendations,”
the proposed rule must provide “an explanation of the reasons
for such differences.” Id. Section 307(d)(6) of the Act
requires that the final promulgated rule must also “be
accompanied by . . . a statement of basis and purpose like that
referred to in paragraph (3) with respect to a proposed rule.”
Id. § 7607(d)(6)(A). Thus if, as here, EPA departs from
CASAC’s recommendations in the final rule, EPA must also
explain there its reasons for doing so. See American Farm
Bureau, 559 F.3d at 521 (concluding that EPA failed in the
final rule “adequately to explain its reason for not accepting
the CASAC’s recommendations”).
Congress intended that CASAC’s expert scientific
analysis aid not only EPA in promulgating NAAQS but also
the courts in reviewing EPA’s decisions. As Congress
explained, CASAC’s “views are to be included in the record
 35
of any such rulemaking proceeding and, therefore, to be
considered by the courts in reviewing the Administrator’s
action or inaction.” H. Rep. No. 95-294, at 182–83. In order
to enable judicial review and to satisfy its statutory obligation
to explain its reasons for departing from CASAC, EPA must
be precise in describing the basis for its disagreement with
CASAC. If EPA’s quarrel is with CASAC’s scientific
analysis, then in order to preserve the integrity of CASAC’s
scientific role, EPA must give a sound scientific reason for its
disagreement. In reviewing such scientific explanations, we
undertake a “searching and careful” inquiry into the facts “to
ascertain whether there is substantial evidence in the record
when considered as a whole which supports the
Administrator’s determinations.” Lead Industries Ass’n, 647
F.2d at 1145–46 (internal quotation marks omitted).
Alternatively, EPA could accept CASAC’s scientific analysis
yet explain the policy considerations that led it to select a
different level than that recommended by CASAC. See id. at
1147. Of course, EPA’s policy judgments “are not susceptible
to the same type of verification or refutation by reference to
the record as are some factual questions,” and thus “our
paramount objective” in reviewing them “is to see whether
the agency, given an essentially legislative task to perform,
has carried it out in a manner calculated to negate the dangers
of arbitrariness and irrationality.” National Lime Ass’n v.
EPA, 627 F.2d 416, 431 n.48 (D.C. Cir. 1980) (internal
quotation marks omitted).

In this case, the CASAC Ozone Review Panel was
composed of twenty-three scientists who are professors,
analysts, and other practitioners in fields such as medicine,
anatomy, environmental science, and chemical engineering.
Drawing on this substantial expertise, the twenty-three
members of the panel, in an October 2006 letter to EPA
following CASAC’s peer review of the second draft of the
 36
agency’s Ozone Staff Paper, unanimously recommended that
“the current primary ozone NAAQS be revised and that the
level that should be considered for the revised standard be
from 0.060 to 0.070 ppm.” Oct. 2006 CASAC Letter, at 5. In
explaining the basis for this recommendation, CASAC noted
that “[a] large body of data clearly demonstrates adverse
human health effects at the current level of the 8-hr primary
ozone standard.” Id. According to CASAC, “[r]etaining this
standard would continue to put large numbers of individuals
at risk for respiratory effects and/or significant impact on
quality of life including asthma exacerbations, emergency
room visits, hospital admissions and mortality.” Id.

CASAC also noted a large body of studies providing
“evidence for adverse health effects at concentrations lower
than the current standard.” Id. at 3. Among this evidence was
a “broad range of epidemiologic and controlled exposure
studies” observing multiple “adverse health effects due to
low-concentration exposure to ambient ozone.” Id. at 4. In
addition, CASAC explained that the Adams (2006) study had
observed “[s]tatistically-significant decrements in lung
function . . . at the 0.08 ppm exposure level,” as well as
“adverse lung function effects . . . in some individuals at 0.06
ppm.” Id. at 3. CASAC also noted that “these findings were
observed in healthy volunteers” and that asthmatics and
children had been found in other studies “to be more sensitive
and to experience larger decrements in lung function in
response to ozone exposures than would healthy volunteers.”
Id. at 4. Finally, pointing to the exposure and risk
assessments, CASAC explained that “a significant decrease in
adverse effects due to ozone exposures can be achieved by
lowering the exposure concentrations below the current
standard,” noting that “[b]eneficial effects in terms of
reduction of adverse health effects were calculated to occur at
the lowest concentration considered (i.e., 0.064 ppm).” Id. On
 37
the basis of all this evidence, CASAC concluded that “the
current primary 8-hr standard of 0.08 ppm needs to be
substantially reduced to be protective of human health,
particularly in sensitive subpopulations” and that the standard
should be set within the range of 0.060 to 0.070 ppm. Id. at 4–
5. CASAC reiterated this recommendation in a March 2007
letter to EPA, underscoring that “overwhelming scientific
evidence” supported its recommendation “that the level of the
current primary ozone standard should be lowered from 0.08
ppm to no greater than 0.070 ppm.” Mar. 2007 CASAC
Letter, at 2.

When EPA issued its notice of proposed rulemaking, it
proposed to revise the primary ozone standard to within a
range from 0.070 ppm, the high end of CASAC’s
recommended range, to 0.075 ppm. 2007 Proposed Rule, 72
Fed. Reg. at 37,878. EPA explained why it believed a
standard set below 0.070 ppm would be inappropriate. Id. at
37,880. In the final rule, EPA departed from CASAC’s
recommended range and set the standard at 0.075 ppm. EPA
acknowledged that this standard was “above the range
recommended by the CASAC.” 2008 Final Rule, 73 Fed. Reg.
at 16,482. In explaining its departure, EPA catalogued its
disputes with CASAC over the interpretation of specific
bodies of scientific evidence and also noted that “the basis for
[CASAC’s] recommendation appears to be a mixture of
scientific and policy considerations.” Id. “[T]here is,” EPA
stated, “no bright line clearly directing the choice of level, and
the choice of what is appropriate is clearly a public health
policy judgment entrusted to the Administrator.” Id. at
16,482–83. In explaining this policy judgment, EPA reasoned
that “[a] standard set at a level lower than 0.075 would only
result in significant further public health protection if, in fact,
there is a continuum of health risks in areas with 8-hour
average O3 concentrations that are well below the
 38
concentrations observed in the key controlled human
exposure studies and if the reported associations observed in
epidemiological studies are, in fact, causally related to O3 at
those lower levels.” Id. at 16,483. “Based on the available
evidence,” EPA declared that it was “not prepared to make
these assumptions.” Id. “Taking into account the uncertainties
that remain in interpreting the evidence from available
controlled human exposure and epidemiological studies at
very low levels,” EPA concluded that “the likelihood of
obtaining benefits to public health with a standard set below
0.075 ppm O3 decreases, while the likelihood of requiring
reductions in ambient concentrations that go beyond those
that are needed to protect public health increases.” Id. EPA
thus “judge[d] that the appropriate balance to be drawn, based
on the entire body of evidence and information available in
this review, is a standard set at 0.075.” Id.

This explanation rests largely on EPA’s policy judgment
about the appropriate NAAQS level. We have explained that,
where EPA operates within the realm of uncertain science, its
decisions about the appropriate NAAQS level must
“necessarily . . . rest largely on policy judgments.” Lead
Industries Ass’n, 647 F.2d 1147 (internal quotation marks
omitted). But this presupposes that the scientific evidence is
actually uncertain—a question that itself requires a scientific
determination. EPA did not make such a specific scientific
determination about the 0.070 ppm level that served as the
ceiling of CASAC’s recommendation; instead, EPA referred
generally to declining certainty below 0.075 ppm. Had
CASAC reached a scientific conclusion that adverse health
effects were likely to occur at the 0.070 ppm level, EPA’s
failure to justify its uncertainty regarding the existence of
 39
adverse health effects at this level would be unacceptable. 6
Indeed, it is a familiar principle that agencies may not
“merely recite the terms ‘substantial uncertainty’ as a
justification for [their] actions”; instead, they “must explain
the evidence which is available, and must offer a rational
connection between the facts found and the choice made.”
State Farm, 463 U.S. at 52 (internal quotation marks omitted).
In other words, EPA must explain why the evidence on which
CASAC relied cannot support the degree of confidence
CASAC placed in it. This is especially true given the added
layer of stringency imposed by EPA’s obligations under
section 307(d)(6).

But we are unable to determine whether CASAC reached
any such scientific conclusion. Although CASAC stated that
“overwhelming scientific evidence” supported its
recommendation that the standard be set no higher than 0.070
ppm, Mar. 2007 CASAC Letter, at 2, it never explained
whether this proposal was based on its scientific judgment
that adverse health effects would occur at that level or instead
based on its more qualitative judgment that the range it
proposed would be appropriately protective of human health
with an adequate margin of safety. Indeed, although CASAC
concluded that “there is no longer significant scientific
uncertainty regarding [its] conclusion that the current 8-hr
primary NAAQS must be lowered,” given the “large body of
data clearly demonstrat[ing] adverse human health effects at
the current level,” CASAC recognized that “[s]cientific
uncertainty does exist with regard to the lower level of ozone
6
This conclusion concerns only disagreements regarding the
certainty of the science; of course, EPA could also have accepted
CASAC’s scientific conclusion and explained its view that any
health effects at that level were not severe enough to be considered
“adverse.”
 40
exposure that would be fully-protective of human health.”
Oct. 2006 CASAC Letter, at 5.

To be sure, EPA’s statutory obligation to respond to
CASAC does not evaporate whenever CASAC exercises
judgment amidst scientific uncertainty. Quite to the contrary,
had CASAC acknowledged uncertainty in the scientific
evidence but explained that, based on its expert scientific
judgment, it nonetheless believed adverse health effects were
likely to occur at the 0.070 ppm level, then section 307(d)(6)
would have required EPA to explain why it disagreed with
this scientific conclusion. Put differently, to the extent that
CASAC has exercised scientific judgment, EPA must respond
in kind. But because CASAC never made clear the precise
basis for its recommendation, all we know for certain is this:
both CASAC and EPA believed the existence of adverse
health effects to be certain at the 0.08 ppm level and reached
differing conclusions about what level below 0.08 ppm was
requisite to protect the public health with an adequate margin
of safety.

The task of determining what standard is “requisite” to
protect the qualitative value of public health or what margin
of safety is “adequate” to protect sensitive subpopulations
necessarily requires the exercise of policy judgment. Here,
EPA’s policy judgment was informed by its view of the
limitations of the scientific evidence—namely, that at lower
levels of ozone exposure, the clinical and epidemiological
studies provide less conclusive evidence of the existence of
adverse health effects. See 2008 Final Rule, 73 Fed. Reg. at
16,483 (noting “the uncertainties that remain in interpreting
the evidence from available controlled human exposure and
epidemiological studies at very low levels”). Striking a
balance between “the increasing uncertainty associated with
[its] understanding of the likelihood of such effects at lower
 41
O3 exposure levels” and “concern about the potential for
health effects and their severity,” id. at 16,477, EPA set the
standard at 0.075 ppm, a level the agency believed to be
“appreciably below” the 0.08 ppm level at which both EPA
and CASAC expressed certainty about the existence of
adverse health effects, id. at 16,483. Absent a definitive
scientific conclusion from CASAC that adverse health effects
would occur at the 0.070 ppm level, we must assume that it
too took these same considerations into account and simply
exercised its judgment to recommend a standard set at a lower
level. Although both CASAC and EPA must exercise public
health policy judgment when confronted with scientific
evidence that does not direct it to a specific outcome, it is to
EPA’s judgment that we must defer.

In our view, this conclusion is perfectly consistent with
the role Congress intended CASAC to play in the NAAQS-
setting process. In order to ensure that EPA’s NAAQS
decisions rest on sound scientific judgment, Congress
required EPA not only to describe CASAC’s
recommendations in any rulemaking but also, if it departs
from such recommendations, to explain its reasons for doing
so. See 42 U.S.C. § 7607(d)(3), (6). But in order for EPA to
explain adequately its reasons for disagreeing with CASAC,
CASAC itself must be precise about the basis for its
recommendations. Because in this case CASAC failed to
specify whether the 0.070 ppm level it recommended as a
maximum rested on a scientific conclusion about the
existence of adverse health effects at that level, EPA’s
invocation of scientific uncertainty and more general public
health policy considerations satisfies its obligations under the
statute.
 42
IV.
We turn finally to EPA’s decision to set the secondary
ozone NAAQS “identical in every way to the revised primary
standard.” 2008 Final Rule, 73 Fed. Reg. at 16,500. The
governmental and environmental petitioners argue, among
other things, that EPA’s failure to “specify a level of air
quality . . . [that] is requisite to protect the public welfare,” 42
U.S.C. § 7409(b)(2), violates the statute as interpreted in our
decision in American Farm Bureau. Because we agree that
EPA’s justification for the secondary standard is inadequate
under American Farm Bureau, we need not reach petitioners’
other arguments.

As described above, the Clean Air Act requires secondary
NAAQS to “specify a level of air quality the attainment and
maintenance of which . . . is requisite to protect the public
welfare from any known or anticipated adverse effects
associated with the presence of such air pollutant in the
ambient air.” Id. Regarding ozone, EPA set the secondary
NAAQS in 1997 to protect against harmful effects on
vegetation and indirect effects on other ecosystem
components. See 2008 Final Rule, 73 Fed. Reg. at 16,485. In
the current review of the secondary standard, before EPA
came to its final decision, agency staff examined new
scientific evidence and risk assessments that evaluated
ozone’s welfare effects. Id. On the basis of this new evidence,
EPA staff concluded that the existing 0.08 ppm 8-hour
standard was inadequate because ozone at that level directly
causes adverse effects to vegetation and has indirect adverse
effects on soil, water, and wildlife. 2007 Proposed Rule, 72
Fed. Reg. at 37,883, 37,897–99. EPA staff also considered
whether the evidence still justified a standard that measured
ozone over an 8-hour interval or whether instead the
secondary standard should measure ozone cumulatively over
a seasonal period. Id. at 37,882–83. In the end, EPA staff
 43
found that new evidence about the cumulative effect of ozone
on vegetation supported a seasonal standard and
recommended that the agency consider a range of seasonal
levels between 7 and 21 ppm-hours. Id. at 37,900, 37,903.

CASAC unanimously agreed with EPA staff that adverse
effects on vegetation occur under the existing standard and
that “it is not appropriate to try to protect vegetation . . . by
continuing to promulgate identical primary and secondary
standards for O3.” 2008 Final Rule, 73 Fed. Reg. at 16,492.
All but one member of the CASAC panel “encourage[d]
[EPA] to establish an alternative cumulative secondary
standard for O3 and related photochemical oxidants that is
distinctly different in averaging time, form and level from the
currently existing or potentially revised 8-hour primary
standard.” 2007 Proposed Rule, 72 Fed. Reg. at 37,899 &
n.62 (internal quotation marks omitted). CASAC also agreed
with EPA staff that the lowest seasonal level that the agency
should consider was 7 ppm-hours, but recommended that
EPA consider a level no higher than 15 ppm-hours. Id. at
37,903.

In the final rule, EPA agreed that new evidence indicates
that ozone causes adverse effects on vegetation and related
ecosystems at the current level of the secondary standard and
concluded that it was appropriate to revise the secondary
standard to provide increased protection. 2008 Final Rule, 73
Fed. Reg. at 16,496, 16,499–500. Regarding the
recommendations to adopt a cumulative seasonal standard,
EPA cited a staff analysis that found “significant overlap”
between counties expected to meet the revised 8-hour primary
standard and counties that would meet a cumulative seasonal
standard. Id. at 16,499. EPA “focused [its] consideration on a
level for an alternative [seasonal] standard at the upper end of
the proposed range (i.e., 21 ppm-hours)” and found
 44
“essentially no counties with air quality that would be
expected both to exceed such an alternative [seasonal]
standard and to meet the revised 8-hour primary standard.” Id.
at 16,499–500. From this comparison, EPA concluded that
merely revising the secondary standard to match the revised
primary standard would “provide a significant degree of
additional protection for vegetation” and that “a [seasonal]
standard would be unlikely to provide additional protection in
any areas beyond that likely to be provided by the revised
primary standard.” Id. at 16,499–500. Citing the “significant
uncertainties in determining or quantifying the degree of risk
attributable to varying levels of O3 exposure, the degree of
protection that any specific cumulative, seasonal standard
would produce, and the associated potential for error in
determining the standard that will provide a requisite degree
of protection,” EPA rejected a cumulative seasonal standard
in favor of a secondary standard that was identical to the
revised primary standard. Id. at 16,500.

In American Farm Bureau, we rejected EPA’s
explanation for setting the fine particulate matter secondary
NAAQS—which protects public welfare from adverse
visibility effects—identical to the primary fine particulate
matter standard. 559 F.3d at 530–31. While the primary
standard measured fine particulate matter levels annually,
EPA staff and CASAC had recommended that the secondary
standard measure fine particulate matter over 4- or 8-hour
periods, suggesting a range of appropriate levels for this
alternatively measured standard. Id. at 528. EPA rejected
these recommendations on the ground that the evidence
supporting the recommended alternative standard was
“limited and uncertain,” instead adopting a secondary
standard that was identical to the primary standard. Id. at 529
(internal quotation marks omitted). In so doing, EPA relied on
a comparison purporting to show that the revised primary
 45
NAAQS would actually provide slightly more visibility
protection than one proposed level of the alternative standard.
Id.

Relying on the statute’s plain language—EPA “shall
specify a level of air quality the attainment and maintenance
of which . . . is requisite to protect the public welfare from
any known or anticipated adverse effects,” 42 U.S.C.
§ 7409(b)(2)—we rejected EPA’s explanation, finding that
EPA must “determine what level of visibility protection is
requisite to protect the public welfare,” American Farm
Bureau, 559 F.3d at 530. We also found that EPA’s reliance
on the comparison between the primary standard and the
recommended secondary standards “fail[ed] on its own
terms.” Id. “[T]wo-thirds of the potential standards within the
CASAC’s recommended range,” we explained, “would be
substantially more protective than the primary standards,” and
“EPA failed to explain why it looked only at one of the few
potential standards that would be less protective.” Id.
Furthermore, we faulted EPA’s failure to respond to technical
problems with the comparison identified by CASAC and EPA
staff. Id. at 530–31.

Although American Farm Bureau was decided after EPA
issued the rule challenged here, the decision is binding on us
now—a proposition that EPA nowhere disputes. Bradley v.
School Board of City of Richmond, 416 U.S. 696, 714 (1974)
(“[A]n appellate court must apply the law in effect at the time
it renders its decision.” (quoting Thorpe v. Housing Authority
of City of Durham, 393 U.S. 268, 281 (1969))). Indeed, the
statutory requirement that the secondary NAAQS “specify a
level of air quality the attainment and maintenance of
which . . . is requisite to protect the public welfare,” 42 U.S.C.
§ 7409(b)(2), existed when EPA issued the rules at issue in
American Farm Bureau and here.
 46
EPA’s explanation for setting the secondary standard
identical to the primary standard fails under American Farm
Bureau. As we explained there, it is insufficient for EPA
merely to compare the level of protection afforded by the
primary standard to possible secondary standards and find the
two roughly equivalent. EPA must expressly “determine what
level of . . . protection is requisite to protect the public
welfare,” American Farm Bureau, 559 F.3d at 530, and
explain why this is so. Here EPA found “significant overlap”
between the revised primary standard and “selected levels” of
a seasonal standard, 2008 Final Rule, 73 Fed. Reg. at 16,499,
and it did say that the revised primary standard “would be
sufficient to protect public welfare from known or anticipated
adverse effects,” id. at 16,500. But it justified this conclusion
only by comparing the revised primary standard to a seasonal
level of 21 ppm-hours that EPA never “specif[ed]” was
“requisite to protect the public welfare,” 42 U.S.C.
§ 7409(b)(2)—exactly what American Farm Bureau held is
inconsistent with the statute.

EPA argues that it “identified a target level of protection
in terms of a cumulative, seasonal standard.” Respondent’s
Br. 122. In support, the agency points to the sentence in the
final rule stating that EPA “focused [its] consideration on a
level . . . at the upper end of the proposed range (i.e., 21 ppm-
hours).” 2008 Final Rule, 73 Fed. Reg. at 16,499–500. But
neither this statement nor anything else EPA said indicated
that the 21 ppm-hours level was “requisite to protect the
public welfare.” Perhaps more importantly, EPA never
explained why a 21 ppm-hours level would, in fact, be
requisite to protect vegetation. That a seasonal standard of 21
ppm-hours was one of the levels proposed by EPA staff
hardly shows that the level was “requisite to protect the public
welfare.”
 47
Also as in American Farm Bureau, EPA’s comparison
between the primary and secondary standards “fails on its
own terms.” 559 F.3d at 530. Although the comparison
between the revised 8-hour standard and a seasonal standard
showed that the level of protection afforded by the revised
primary standard would be arguably equivalent to the level of
protection afforded by a 21 ppm-hours seasonal standard, the
comparison also showed that the primary standard would
offer less protection than other seasonal levels within the
range recommended by CASAC and EPA staff. See 2007
Proposed Rule, 72 Fed. Reg. at 37,892–93. In American Farm
Bureau, “EPA failed to explain why it looked only at one of
the few potential standards that would be less protective . . .
than the primary standard,” 559 F.3d at 530; in this case, EPA
failed to explain why it looked only at one potential seasonal
standard that the primary standard would arguably protect as
well as.

At oral argument, counsel for EPA repeatedly insisted
that petitioners “tacitly conceded” that the agency identified a
target level of protection by “criticiz[ing] the reason EPA
focused on 21 [ppm-hours].” Oral Arg. Rec. 1:50:43–51:06;
see also id. 1:50:02–16. But counsel confuses assuming a
premise for the sake of argument with conceding the point.
Petitioners argue both that EPA failed to identify a target level
of protection and that, even if EPA had in fact determined that
the “requisite” level was 21 ppm-hours, that finding was
irrational. See Environmental Petitioners’ Br. 35–37 (“EPA
Acted Illegally and Arbitrarily in Failing to Identify the Level
of Air Quality Requisite to Protect Against Adverse
Vegetation Impacts.”); id. at 37–39 (“EPA’s Decision on the
Secondary Standard Was Irrational.”); id. at 39–40 (“EPA’s
Attempts to Justify Its Secondary Standard Were
Groundless.”). Contending that settling on 21 ppm-hours
 48
would be senseless hardly precludes petitioners from arguing
that EPA never expressly made the required determination.

Because EPA failed to determine what level of protection
was “requisite to protect the public welfare,” EPA’s
explanation for the secondary standard violates the Act. We
therefore remand this portion of the final rule for further
explanation or reconsideration by EPA. In the meantime, we
leave the standard in place rather than vacating the rule.
“First, the EPA’s failure adequately to explain itself is in
principle a curable defect. Second, vacating a standard
because it may be insufficiently protective would sacrifice
such protection as it now provides, making the best an enemy
of the good.” American Farm Bureau, 559 F.3d at 528. Given
these principles, neither EPA nor petitioners advocate vacatur.

V.
For the foregoing reasons, we remand the secondary
NAAQS to EPA for reconsideration in view of this opinion.
In all other respects, the petitions for review are denied.

So ordered.