In March 1987, Dr. Frank E. Young, the Commissioner of the Food and Drug Ad*214ministration (“FDA”), issued a final ruling certifying as safe for human consumption the color additive FD & C Blue No. 2. The Health Research Group, the Center for Science in the Public Interest, and Nancy Hendree Simpson (“petitioners”) contend that an animal study upon which the FDA relied to certify the safety of the food additive is invalid. We hold that a fair evaluation of the record as a whole supports the FDA’s determination that Blue No. 2 was adequately tested and shown to be safe for human consumption. Accordingly, we affirm.
I. Background
A.
Under the 1960 Color Additive Amendments to the Federal Food, Drug, and Cosmetic Act (“FDC Act”), the Secretary of the Department of Health and Human Services is responsible for “listing” color additives as “suitable and safe” for uses in food, drugs, or cosmetics. 21 U.S.C. § 376(b)(1) (1982). The authority to “list” an additive is delegated to the FDA Commissioner by regulation. See 21 C.F.R. § 5.10 (1988). The FDA may list a color additive if the data presented to the agency establish that the substance is safe for the use proposed. FDA regulations state that a color additive is “safe” if “there is convincing evidence that establishes with reasonable certainty that no harm will result from the intended use of the color additive.” 21 C.F.R. § 70.3(i) (1988). Under the provision known as the “Delaney Clause,” a food dye may not be listed as safe “if the additive is found by the Secretary to induce cancer when ingested by man or animal.” 21 U.S.C. § 376(b)(5)(B) (1982).
On July 12, 1960 — the effective date of the Color Additive Amendments — all color additives then on the market, including Blue No. 2, were deemed “provisionally listed” pending completion of studies to evaluate their safety. In March 1968, the Certified Color Industry Committee (now the Certified Color Manufacturers’ Association, Inc. (“CCMA”)), petitioned the FDA to commence procedures for the final listing of Blue No. 2 for use in food and drugs. The notice of the filing was published in the Federal Register, 33 Fed.Reg. 9,627 (July 2, 1968). The FDA extended the provisional listing of Blue No. 2 a number of times from 1968 to 1981 because of inadequate data. In 1981, CCMA submitted its final reports of animal feeding studies conducted on Blue No. 2. These CCMA studies included a long-term feeding study in mice and a similar study in rats. Bio/dynamics, Inc. conducted the studies for CCMA in accordance with protocols approved by the FDA.
After scrutinizing these data, the FDA published its final rule permanently listing the dye as “safe” on the basis of the results of the long-term carcinogenicity studies. 48 Fed.Reg. 5,252 (Feb. 4, 1983). Petitioners filed objections to the FDA’s decision claiming, among other things, that the concentration of Blue No. 2 administered in the CCMA rat study to those animals receiving the highest dosage was too low. According to a manual published in 1982 by the FDA Bureau of Foods known as the “Red Book,” a carcinogenesis study should be designed so that the highest concentration of a suspected carcinogen that is fed to test animals is the so-called “maximum tolerated dose” (“MTD”) for that substance. See Toxicological Principles for the Safety Assessment of Direct Food Additives and Color Additives Used in Food (“Red Book”) at 57-58 (1982), Joint Appendix (“J.A.”) at 83-84. The MTD of a chemical additive is described in the Red Book as the lowest concentration of dye to produce signs of toxicity without increasing mortality or decreasing longevity. See id. at 58, J.A. at 84. Petitioners contend that the two percent dosage administered to the animals receiving the highest concentration of dye in the CCMA rat study is not the MTD because those animals did not display the physiological effects that are the accepted indicators of MTD toxicity.
In April 1984, an Administrative Law Judge (“ALJ”) conducted an evidentiary hearing during which experts representing disparate views as to the validity of the CCMA study testified. The FDA’s Center *215for Food Safety and Applied Nutrition (“Food Center”) and the CCMA argued for approval of the dye. Petitioners in this case appeared before the ALJ to oppose the listing. In February 1985, the ALJ decided that the CCMA study was reliable and that two percent was a proper MTD. The FDA Commissioner affirmed the ALJ’s decision two years later. Petitioners now petition this court for reversal of the FDA’s decision. Their sole contention on appeal is that the MTD in the CCMA rat study was set too low, and that the study is therefore inadequate to demonstrate the safety of Blue No. 2 dye for human consumption.
B.
An understanding of the dispute concerning the MTD selected in this case requires some familiarity with the FDA’s general protocol for carcinogenicity studies. Standards for long-term animal studies to evaluate cancer-causing potential are set forth in the Red Book. Supplementary discussions on particular aspects of carcinogenicity testing are scattered among various government reports and documents presented as part of the record in this rulemaking. Because tumors take a long time to develop at the small doses of color additives contained in ordinary food, studies are conducted using higher doses for a shorter period. The results are extrapolated to estimate the risk of cancer. According to the Red Book, test animals receive the dye in two phases. First, parent animals ingest the substance prior to mating, and the mother continues to eat feed mixed with dye during pregnancy and the nursing of her offspring. The period during which the progeny — or “second generation” pups — are exposed to the test substance from the mother before birth and during nursing is called the in útero phase. The adult phase of the study begins at weaning. Pups selected as controls receive a normal diet. The remaining animals are divided into three “experimental” groups. Each group ingests food impregnated with dye at one of three graduated concentrations throughout the duration of the study (about 30 months).
According to the Red Book, the highest of the three doses should be the MTD for the substance tested. The lower concentrations are determined as standard percentages of the highest dose. The Red Book defines the MTD as a dose “sufficiently high to elicit signs of minimal toxicity without substantially altering the normal lifespan due to effects other than tumors.” Red Book at 58, J.A. at 84. Excerpts in the record from reports published by the Office of Science and Technology Policy (“OSTP Report”) and the National Toxicology Program Board of Scientific Advisors (“NTP Report”) contain substantially similar general definitions of the MTD. The only additional assertion in the Red Book concerning specific criteria for the MTD states that “[sjigns of toxicity are those that may be indicated by alterations in serum enzyme levels or slight depression of body weight gain (less than 10 percent).” Red Book at 58, J.A. at 84.
Proper determination of the MTD is crucial to the validity of a study to evaluate the cancer-causing potential of a food additive. If the highest dose fed to experimental animals is lower than the “true” MTD, it may produce a false-negative result: the dose fed to the animals may not cause tumor formation, whereas a properly selected MTD might have elicited this response. Thus, the validity of the MTD chosen for the CCMA study is central to the question of whether the FDA was justified in relying on that study and in listing Blue No. 2 as safe under the FDC Act.
C.
In explaining his conclusion that the two percent MTD chosen for the CCMA rat feeding study was correct, the ALJ considered a variety of evidence. He extensively discussed the FDA’s reliance on a rat feeding study conducted in 1966 by Hansen and colleagues (“Hansen study”) and petitioners’ objections to the agency’s use of these data. The study was a conventional, multiple-dose rat feeding study using a sample size too small to test carcinogenicity, but large enough to indicate the MTD. The Hansen study showed that the *216experimental rats — which Hansen and his colleagues exposed to the dye during the adult phase only — failed to gain weight normally at both the two percent and five percent feeding levels, but that the effect was not statistically significant at two percent. At least one of petitioners’ expert witnesses testified that the Hansen results, combined with other data, failed to show that two percent was the proper MTD for Blue No. 2. Opposing experts testified that the study was consistent with an MTD of two percent. Some FDA experts explained that differences in the design of the Hansen and CCMA studies — in particular, the lack of in útero exposure in the Hansen study — coupled with independent observations of toxicity qnd mortality at the close of the in útero phase of the CCMA study, supported setting the MTD “at or near” two percent.
In rejecting petitioners’ contention that the Hansen study did not demonstrate a two percent MTD, the AU relied on the latter testimony and also cited additional evidence. This included a 1974 study in which mice developed anemia at a 1.6 percent dose level, and a 1969 short-term pig toxicity study that revealed anemia at a dose corresponding to two percent Blue No. 2. The AU rejected the petitioners’ argument that these studies were irrelevant because anemia alone, without reduction in weight gain, is not an accepted sign of MTD toxicity. The AU explained that, as “the [petitioner] itself has asserted[,] * * * slight reduction in weight gain * * * is by no means the only indicator that the MTD has been achieved.” AU Decision at 23, J.A. at 48 (citations and internal quotes omitted).
The FDA Commissioner (“Commissioner”) affirmed the AU’s decision and adopted the AU’s findings “to the extent that [they are] consistent with” the Commissioner’s discussion. The Commissioner did not specifically address the petitioners’ objections to reliance on the Hansen study in the FDA’s original published listing, or the AU’s conclusions regarding the expert testimony concerning the Hansen study. The Commissioner instead concentrated on observations from the main CCMA study bearing on the validity of the determination of the MTD as two percent. The FDA acknowledged the petitioners’ arguments that the “classic” signs of toxicity mentioned in the Red Book — alteration in serum enzyme levels and significant depression in body weights — were not observed in the adult male rats fed the two percent dose in the CCMA study. However, the FDA agreed with the AU that the absence of these “classic” indicia was not dispositive because a “broad range of biological information * * * [including] alterations in body and organ weight and clinically significant changes in hematologic, urinary, and clinical chemistry measurements” could be used to assess the MTD. FDA Decision at 11, J.A. at 11. The FDA noted that “[t]he long term [CCMA] rat study showed increased spleen and liver weights and lower fasting glucose levels in the FI [i.e. second generation] mature adults. There was also a reduction in body weight of the treated pups by approximately 15 percent at weaning [and p]up survival decreased in the mid and high dose groups.” The Commissioner concluded that “these adverse effects are sufficient evidence that the rat feeding study was conducted at doses of Blue No. 2 at or near the MTD.” Id.
II. Discussion
A.
The FDC Act provides that “[t]he findings of the [FDA] with respect to questions of fact shall be sustained if based upon a fair evaluation of the entire record at such hearing.” 21 U.S.C. § 348(g)(2) (1982). In determining the scope of review, Congress responded to the industry's concern that a more deferential standard might allow agency officials to credit the testimony of selected scientists while disregarding “contradictory evidence of possibly equal or even greater substance.” S.Rep. No. 2422, 85th Cong., 2d Sess. 8 (1958), U.S.Code Cong. & Admin.News 1958, pp. 5300, 5307. Congress indicated that the court must determine whether the secretary has “given appropriate consideration to the inferences which should fairly be drawn from all of *217the evidence of the record.” Id. at 9, U.S. Code Cong. & Admin.News 1958, p. 5308.
Although Congress may have intended to create a higher standard of review than the traditional “substantial evidence” standard, it is far from clear that an evaluation by this court of the agency’s decision on the record as a whole diverges significantly from substantial evidence review. The Supreme Court in Universal Camera Corp. v. NLRB, 340 U.S. 474, 488, 71 S.Ct. 456, 464, 95 L.Ed. 456 (1951), observed that “[t]he substantiality of evidence must take into account whatever in the record fairly detracts from its weight. This is clearly the significance of the requirement * * * that the courts consider the whole record.” See also 5 K. Davis, Administrative Law Treatise § 29.22, at 437-38 (2d ed. 1984) (quoting Jaffe, Administrative Procedure Re-examined: The Benjamin Report, 56 Harv.L.Rev. 704, 733 (1943) (“Obviously responsible men would not exercise their judgment on only that part of the evidence that looks in one direction.”))
Whatever the “fair evaluation” standard of review, as opposed to the traditional review standard, signifies in the abstract, we think our mandate in this case is clear. We must uphold the FDA’s decision if it reveals that significant evidence on both sides of the question has been considered and that the agency has explained its conclusions in light of significant objections. Having said this, we note that the fair evaluation standard does not license this court to conduct a de novo review in which this court substitutes its judgment concerning the evidence for that of the administrative decisionmaker. See Consolo v. Federal Maritime Comm’n, 383 U.S. 607, 619 n. 17, 86 S.Ct. 1018, 1026 n. 17, 16 L.Ed.2d 131 (1966) (de novo review is not presumed without specific statutory authorization).
A re-weighing by this court of the mass of scientific evidence presented to the FDA would not only be inappropriate as a matter of law, but impossible as a practical matter. “The stringency of our review in a given case” must unavoidably depend upon “the ability of the court effectively to evaluate the questions posed.” Natural Resources Defense Council v. SEC, 606 F.2d 1031, 1050 (D.C.Cir.1979). This court has “neither the expertise nor the resources to evaluate complex scientific claims.” Thompson Medical Co. v. FTC, 791 F.2d 189, 196 (D.C.Cir.1986), cert. denied, 479 U.S. 1086, 107 S.Ct. 1289, 94 L.Ed.2d 146 (1987). Our inquiry requires “searching and careful” consideration of whether the agency ignored highly relevant evidence or formed a conclusion for which record support is absent or clearly inadequate to the commonsense observer. See Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 416, 91 S.Ct. 814, 823, 28 L.Ed.2d 136 (1971). However, we cannot decide which highly trained expert to believe, or whether technical evidence beyond our ken supports the proposition it is asserted to support.
Petitioners argue that the relevant evidence can be easily understood by a non-scientist and its evaluation requires only “common sense.” Some of petitioners’ arguments and some of the evidence before the FDA can be evaluated by non-experts, and we adjust the intensity of our scrutiny accordingly. However, the proper resolution in this case of a number of disputed points concerning the MTD requires sophisticated scientific judgment. Deference to agency expertise is especially appropriate with respect to the issue of selection of the MTD — an area characterized, as we shall see, by imprecision, substantial controversy, and evolving standards. See Weyerhaeuser Co. v. Costle, 590 F.2d 1011, 1025-26 (D.C.Cir.1978).
Petitioners also fault the FDA’s terse final decision upholding the listing of Blue No. 2 for its alleged failure to address each “significant contention[] made by * * * participants]” as required by pertinent regulations. 21 C.F.R. § 12.120(b)(3) (1988); see id. § 12.130(c). We note that neither the applicable regulations nor the Administrative Procedure Act requires separate, specific rulings on each exception to a decision. The pertinent regulation speaks of “significant” objections. Thus, the FDA is not required to address every argument advanced by petitioners at the *218hearing no matter how minor or inconsequential the argument may be. The agency need only state the main reasons for its decision and indicate that it has considered the most important objections. See ASG Industries, Inc. v. CPSC, 593 F.2d 1323, 1329 (D.C.Cir.), cert. denied, 444 U.S. 864, 100 S.Ct. 133, 62 L.Ed.2d 87 (1979); Kennecott Copper Corp. v. EPA, 612 F.2d 1232, 1236 (10th Cir.1979). As we discuss below, the FDA decision on Blue No. 2 satisfies these requirements.
B.
Petitioners contend that the FDA relied on deficient evidence in determining that the MTD was administered in the CCMA rat study. Petitioners make five main objections, which we discuss in turn.
1. Failure to perform a pilot study
One way in which the FDA erred, petitioners maintain, was by approving the selection of the MTD level without requiring that the CCMA submit the results of a separate, subchronic pilot study. The Red Book “strongly recommend[s]” that the MTD be selected by conducting a “pilot” subchronic range-finding study. See Red Book at 58, 120, J.A. at 84, 87. These studies are shorter in duration and use fewer animals than the full-fledged carcinogenicity studies, but are adequate to indicate the lowest dose at which minimal toxicity appears. Although petitioners concede that Red Book assertions are not binding on the agency, they contend that this document sets forth the agency’s internal guidelines and that these guidelines are in accordance with “generally accepted standards.”
CCMA’s failure to perform a subchronic pilot study is not fatal to the FDA’s ruling. Dr. Jackson, Chief of the Color and Cosmetics Evaluation Branch of the FDA Food Center’s Division of Toxicology offered uncontested testimony that a subchronic range-finding study is the procedure of choice when other sources, such as previously published studies or the full long-term carcinogenicity study itself, provide no usable information — a situation that the FDA found did not obtain for Blue No. 2. One of petitioners’ own expert witnesses admitted that the results of a long-term carcinogenicity study may indicate the MTD with sufficient clarity to obviate the need for a pilot study. It is clear from the record that the pilot study is simply one accepted and efficient method to determine the MTD to be used in the main study, not an iron-clad prerequisite to the validity of the MTD actually selected. The FDA found that effects observed in the main CCMA study revealed the proper MTD. Thus, the FDA justifiably rejected petitioners’ argument that a pilot study was necessary to determine the Blue No. 2 MTD.
2. Choice of physiological signs to indicate the MTD
Petitioners also assert that the FDA arbitrarily departed from accepted standards by accepting changes in spleen and liver weights and depression of blood glucose levels in the second generation adult rats as evidence that two percent was the correct MTD. Petitioners contend that weight loss, enzyme level shift, and shortened lifespan are the exclusive criteria for determining the MTD. They cite the Red Book statements that the MTD should not reduce longevity and that signs of toxicity “may be indicated by changes in serum enzyme levels or slight depression of body weight gain (less than 10 percent).” Red Book at 58, J.A. at 84 (emphasis added).
The FDA’s assertion in its decision that the MTD may be based on effects other than those mentioned in the Red Book is supported by permissive language in the manual coupled with considerable evidence from other sources. We note at the outset that record evidence reveals significant uncertainty in the scientific community with respect to the best method of selecting the MTD. See Office of Science and Technology Policy, Chemical Carcinogens: A Review of the Science and Its Associated Principles (“OSTP Draft Report”), 49 Fed. Reg. 21,594, 21,633 (May 22, 1984), J.A. at 118 (“The question of what maximum dose should be administered may be the most controversial issue concerning bioassays.”); *219see also OSTP Final Report, 50 Fed.Reg. 10,372, 10,413 (Mar. 14, 1985) (same). A number of the scientific experts testifying at the hearing discussed the difficulties with the MTD as a toxicologic concept and the lack of consensus as to the precise criteria — both quantitative and qualitative —for determination of the MTD.
With respect to specific criteria used to assess the MTD, Dr. John Doull, a toxicologist on the Board of Scientific Counselors of the National Toxicology Program, stated that relevant effects could include “any adverse phenomenon” which is detrimental to the health and survival of the exposed population. The OSTP Final Report reiterates that a range of biological effects may be considered:
[While t]he MTD was initially based on a weight gain decrement observed in the subchronic study * * * [m]ore recent studies * * * indicate refinement of MTD selection on the basis of a broader range of biological information. Alterations in body and organ weight and clinically significant changes in hematologic, urinary and clinical chemistry measurements may be useful in conjunction with the usually more definitive toxic, gross, or histopathologic endpoints.
OSTP Final Report, 50 Fed.Reg. at 10,413. On the basis of this statement, published after the promulgation of the Red Book, the FDA was entitled to reject as outmoded exclusive reliance on criteria listed in the FDA manual. The three expert witnesses testifying for petitioners cited no other authority for their assertions that increased mortality and reduction in weight gain were the definitive measures of the MTD. The FDA plainly chose to discount these experts’ opinions as superseded by more recent evidence. It is the prerogative of the FDA to weigh and evaluate this conflicting testimony, and the record indicates that the FDA’s conclusions were reasonable. Thus, the FDA’s determination that organ weight and blood glucose effects were valid signs of the MTD is supported by a fair evaluation of the record as a whole.
3. Magnitude of the effects in admit rats
Petitioners argue that, even conceding that the data from the CCMA study are relevant to whether two percent is the MTD in rats for Blue No. 2, the data do not show the degree of effect on parameters that would indicate the MTD was achieved. With respect to liver, spleen, and blood glucose observations, petitioners first argue that the differences supposedly observed between control groups and groups treated for various periods with the highest dose of Blue No. 2 are illusory. In making their elaborate and highly technical argument, petitioners rely on non-statistical comparisons of organ and glucose measurements on particular animals. Petitioners’ contentions are directly contradicted by the evidence on the record, which incontrovertibly shows statistically significant differences in these parameters between groups of control animals and some groups of animals at the highest dose after various periods of exposure. As the government explains in its brief, statistical significance cannot be discerned by the comparison that petitioners make among individual animals, but only by a comparison of averages according to accepted statistical methods. We must defer to the FDA’s explication of the proper method for evaluating statistical significance, which, in any event, accords with well-known principles of statistical analysis.
Petitioners also point to the fact that only some groups of animals fed the highest dose of dye showed the organ effects, that the effects appeared only after several months, and that some of the effects were equivocal. We are aware of no evidence on the record that objectively buttresses these claims that the data are inadequate. Petitioners presented no authority for a rigid requirement that differences be shown in all groups of animals after the shortest period of exposure. At least one FDA expert testified otherwise. Hence, the FDA was entitled to conclude that the magnitude and pattern of the effects reported were sufficient to indicate the MTD.
Moreover, it is clear from the record that it is impossible to measure precisely the *220dosage level at which a color additive begins to exert a toxic effect. The OSTP Report, in both draft and final versions, states that the current methodology does not permit determination of the MTD with exactitude. See 49 Fed.Reg. at 21,633; 50 Fed.Reg. at 10,413. The NTP Report repeats the observation that the MTD is an approximation. See Draft Report, Ad Hoc Panel on Chemical Carcinogenesis Testing and Evaluation of the National Toxicology Program, Board of Scientific Counselors 180 (Feb. 15, 1984), J.A. at 167. Thus, the record does not require us to conclude that more clear-cut data are necessary to determine the MTD. We therefore defer to the FDA’s decision to rely on the organ weight and glucose data.
4. Single-dose study design
Petitioners contend that the reduction in weight gain and increased mortality observed at the end of the nursing phase of the CCMA study in animals fed the two percent dose fails to justify the conclusion that two percent is the MTD for the entire study. Petitioners’ objection focuses on the design of the study. The weight reduction and pup mortality observations, they argue, demonstrate only that two percent is the best maximum dose for the in útero phase, but have no bearing on whether two percent is the appropriate MTD for the adult phase. Petitioners advert to the Red Book statement that “as a result of maternal and fetal toxicity it is often necessary to use lower doses during the in útero phase of the study in order to produce sufficient offspring for the post-weaning phase.” Red Book at 121, J.A. at 87. There is confusing evidence in the record concerning the superiority and reliability of single-dose studies as compared to those in which the dose is elevated at mid-point. We do not deny that a more direct explanation of why the FDA accepted the single-dose design for the CCMA study would have been desirable. However, the short answer to petitioners’ objection is that independent observations in adult rats, already discussed, validate two percent as the MTD for the adult phase. Thus, record evidence justifies a conclusion that raising the highest dose above two percent after the in útero phase would not have resulted in an acceptable study.
5. Significance of the Hansen study
Petitioners contest the ALJ’s reliance on expert testimony that interpreted the Hansen study as indicating that the two percent MTD was valid. The FDA contends that there would remain ample evidence in the record from the results of the CCMA study itself to support the conclusion as to the MTD even if results from the previous animal studies were disregarded. We have been pointed to no evidence contradicting the agency’s assertion that it was entitled to rely exclusively on the adult animal effects expressly noted in the Commissioner’s final decision. Thus, although we think that the AU’s conclusions based on the Hansen study were supported by the record, even if we were to find otherwise, we need not discuss in detail petitioners’ arguments regarding the significance of the Hansen study.
III. Conclusion
We have considered and rejected petitioners’ arguments that the FDA’s approval of FD & C Blue No. 2 food dye as safe for human consumption was not supported by evidence in the record as a whole. In particular, we find reasonable the FDA’s conclusion that experimental animals in a carcinogenicity study submitted to prove the safety of Blue No. 2 received the correct dose of the dye. The arguments not specifically discussed herein were duly considered and found unpersuasive. Accordingly, we affirm the FDA’s decision to list Blue No. 2 food dye as safe for human use.
Affirmed.