with whom The ChieF Justice and Justice Scalia join, dissenting.
In 1973, this Court struck down an Act of the Texas Legislature that had been in effect since 1857, thereby rendering unconstitutional abortion statutes in dozens of States. Roe v. Wade, 410 U. S. 113, 119. As some of my colleagues on the Court, past and present, ably demonstrated, that decision was grievously wrong. See, e. g., Doe v. Bolton, 410 U. S. 179, 221-223 (1973) (White, J., dissenting); Roe v. Wade, supra, at 171-178 (Rehnquist, J., dissenting). Abortion is a unique act, in which a woman’s exercise of control over her own body ends, depending on one’s view, human life or potential human life. Nothing in our Federal Constitution deprives the people of this country of the right to determine whether the consequences of abortion to the fetus and to society outweigh the burden of an unwanted pregnancy on the mother. Although a State may permit abortion, nothing in the Constitution dictates that a State must do so.
In the years following Roe, this Court applied, and, worse, extended, that decision to strike down numerous state statutes that purportedly threatened a woman’s ability to obtain an abortion. The Court voided parental consent laws, see Planned Parenthood of Central Mo. v. Danforth, 428 U. S. 52, 75 (1976), legislation requiring that second-trimester abortions take place in hospitals, see Akron v. Akron Center for Reproductive Health, Inc., 462 U. S. 416, 431 (1983), and even a requirement that both parents of a minor be notified before their child has an abortion, see Hodgson v. Minnesota, 497 U. S. 417, 455 (1990). It was only a slight exaggeration when this Court described, in 1976, a right to abortion “without interference from the State.” Danforth, supra, at 61. The Court’s expansive application of Roe in this period, even more than Roe itself, was fairly described as the “unrestrained imposition of [the Court’s] own, extraconstitutional value preferences” on the American people. Thornburgh v. *981American College of Obstetricians and Gynecologists, 476 U. S. 747, 794 (1986) (White, J., dissenting).
It appeared that this era of Court-mandated abortion on demand had come to an end, first with our decision in Webster v. Reproductive Health Services, 492 U. S. 490 (1989), see id., at 557 (Blackmun, J., concurring in part and dissenting in part) (lamenting that the plurality had “diseard[ed]” Roe), and then finally (or so we were told) in our decision in Planned Parenthood of Southeastern Pa. v. Casey, 505 U. S. 833 (1992). Although in Casey the separate opinions of The CHIEF Justice and Justice Scalia urging the Court to overrule Roe did not command a majority, seven Members of that Court, including six Members sitting today, acknowledged that States have a legitimate role in regulating abortion and recognized the States’ interest in respecting fetal life at all stages of development. See 505.U. S., at 877 (joint opinion of O’Connor, Kennedy, and Souter, JJ.); id., at 944 (Rehnquist, C. J., joined by White, Scalia, and Thomas, JJ., concurring in judgment in part and dissenting in part); id., at 979 (Scalia, J., joined by Rehnquist, C. J., and White and Thomas, JJ., concurring in judgment in part and dissenting in part). The plurality authored by Justices O’Con-nor, Kennedy, and Souter concluded that prior case law “went too far” in “undervalu[ing] the State’s interest in potential life” and in “striking down ... some abortion regulations which in no real sense deprived women of the ultimate decision.” Id., at 875.1 Roe and subsequent eases, according to the plurality, had wrongly “treat[ed] all governmental attempts to influence a woman’s decision on behalf of the potential life within her as unwarranted,” a treatment that was “incompatible with the recognition that there is a substantial state interest in potential life throughout pregnancy.” Id., at 876. Accordingly, the plurality held that so *982long as state regulation of abortion furthers legitimate interests — that is, interests not designed to strike at the right itself — the regulation is invalid only if it imposes an undue burden on a woman’s ability to obtain an abortion, meaning that it places a substantial obstacle in the woman’s path. Id., at 874, 877.
My views on the merits of the Casey plurality have been fully articulated by others. Id., at 944 (Rehnquist, C. J., concurring in judgment in part and dissenting in part); id., at 979 (Scalia, J., concurring in judgment in part and dissenting in part). I will not restate those views here, except to note that the Casey plurality opinion was constructed by its authors out of whole cloth. The standard set forth in the Casey plurality has no historical or doctrinal pedigree. The standard is a product of its authors’ own philosophical views about abortion, and it should go without saying that it has no origins in or relationship to the Constitution and is, consequently, as illegitimate as the standard it purported to replace. Even assuming, however, as I will for the remainder of this dissent, that Casey’s fabricated undue-burden standard merits adherence (which it does not), today’s decision is extraordinary. Today, the Court inexplicably holds that the States cannot constitutionally prohibit a method of abortion that millions find hard to distinguish from infanticide and that the Court hesitates even to describe. Ante, at 923. This holding cannot be reconciled with Casey’s undue-burden standard, as that standard was explained to us by the authors of the plurality opinion, and the majority hardly pretends otherwise. In striking down this statute — which expresses a profound and legitimate respect for fetal life and whieh leaves unimpeded several other safe forms of abortion — the majority opinion gives the lie to the promise of Casey that regulations that do no more than “express profound respect for the life of the unborn are permitted, if they are not a substantial obstacle to the woman’s exercise of the right to choose” whether or not to have an abortion. 505 *983U. S., at 877. Today’s decision is so obviously irreconcilable with Casey’s explication of what its undue-burden standard requires, let alone the Constitution, that it should be seen for what it is, a reinstitution of the pre-Webster abortion-on-demand era in which the mere invocation of “abortion rights” trumps any contrary societal interest. If this statute is unconstitutional under Cctsey, then Casey meant nothing at all, and the Court should candidly admit it.
To reach its decision, the majority must take a series of indefensible steps. The majority must first disregard the principles that this Court follows in every context but abortion: We interpret statutes according to their plain meaning, and we do not strike down statutes susceptible of a narrowing construction. The majority also must disregard the very constitutional standard it purports to employ, and then displace the' considered judgment of the people of Nebraska and 29 other States. The majority’s decision is lamentable, because of the result the majority reaches, the illogical steps the majority takes to reach it, and because it portends a return to an era I had thought wé had at last abandoned.
I
In the almost 30 years since Roe, this Court has never described the various methods of aborting a second- or third-trimester fetus. From reading the majority’s sanitized description, one would think that this ease involves state regulation of a widely accepted routine medical procedure. Nothing could be further from the truth. The most widely used method of abortion during this stage of pregnancy is so gruesome that its use can be traumatic even for the physicians and medical staff who perform it. See App/ 656 (testimony of Dr. Boehm); W. Hern, Abortion Practice 134 (1990). And the particular procedure at issue in this case, “partial birth abortion,” so closely borders on infanticide that 30 States have attempted to ban it. I will begin with a discussion of the methods of abortion available to *984women late in their pregnancies before addressing the statutory and constitutional questions involved.2
1. The primary form of abortion used at or after 16 weeks’ gestation is known as “dilation and evacuation” or “D&E.” 11 F. Supp. 2d 1099, 1103, 1129 (Neb. 1998). When performed during that stage of pregnancy, the D&E procedure requires the physician to dilate the woman’s cervix and then extract the fetus from her uterus with forceps. Id., at 1103; App. 490 (American Medical Association (AMA), Report of the Board of Trustees on Late-Term Abortion). Because of the fetus’ size at this stage, the physician generally removes the fetus by dismembering the fetus one piece at a time.3 11 F. Supp. 2d, at 1103-1104. The doctor grabs a fetal extremity, such as an arm or a leg, with forceps and “pulls it through the cervical os . . . tearing . . . fetal parts from the fetal body ... by means of traction.” Id., at 1104. See App. 55 (testimony of Dr. Carhart). In other words, the physician will grasp the fetal parts and “basically tear off pieces of the fetus and pull them out.” Id., at 267 (testimony of Dr. Stubblefield). See also id., at 149 (testimony of *985Dr. Hodgson) (“[Y]ou grasp the fetal parts, and you often don’t know what they are, and you try to pull it down, and its . . . simply all there is to it”). The fetus will die from blood loss, either because the physician has separated the umbilical cord prior to beginning the procedure or because the fetus loses blood as its limbs are removed. Id., at 62-64 (testimony of Dr. Carhart); id., at 151 (testimony of Dr. Hodgson).4 When all of the fetus’ limbs have been removed and only the head is left in útero, the physician will then collapse the skull and pull it through the cervical canal. Id., at 106 (testimony of Dr. Carhart); id., at 297 (testimony of Dr. Stubblefield); Causeway Medical Suite v. Foster, 43 P. Supp. 2d 604,608 (ED La. 1999). At the end of the procedure, the physician is left, in respondent’s words, with a “tray full of pieces.” App. 125 (testimony of Dr. Carhart).
2. Some abortions after the 15th week are performed using a method of abortion known as induction. 11P. Supp. 2d, at 1108; App. 492 (AMA, Report of the Board of Trustees on Late-Term Abortion). In an induction procedure, the amniotic sac is injected with an abortifacient such as a saline solution or a solution that contains prostaglandin. 11 P. Supp. 2d, at 1108. Uterine contractions typically follow, causing the fetus to be expelled. Ibid.
3. A third form of abortion for use during or after 16 weeks’ gestation is referred to by some medical professionals as “intact D&E.” There are two variations of this method, both of which require the physician to dilate the woman’s cervix. Gynecologic, Obstetric, and Related Surgery 1043 (D. Nichols & D. Clarke-Pearson eds., 2d ed. 2000); App. 271 (testimony of Dr. Stubblefield). The first variation is used only in vertex presentations, that is, when the fetal head is presented first. To perform a vertex-presentation intact D&E, the doctor will insert an instrument into the fetus’ *986skull while the fetus is still in útero and remove the brain and other intracranial contents. 11 F. Supp. 2d, at 1111; Gynecologic, Obstetric, and Related Surgery, supra, at 1043; App. 271 (testimony of Dr. Stubblefield). When the fetal skull collapses, the physician will remove the fetus.
The second variation of intact D&E is the procedure commonly known as “partial birth abortion.”5 11 F. Supp. 2d, at 1106; Gynecologic, Obstetric, and Related Surgery, supra, at 1043; App. 271 (testimony of Dr. Stubblefield). This procedure, which is used only rarely, is performed on mid- to late-second-trimester (and sometimes third-trimester) fetuses.6 Although there are variations, it is generally per*987formed as follows: After dilating the cervix, the physician will grab the fetus by its feet and pull the fetal body out of the uterus into the vaginal cavity. 11 F. Supp. 2d, at 1106. At this stage of development, the head is the largest part of the body. Assuming the physician has performed the dilation procedure correctly, the head will be held inside the uterus by the woman’s cervix. Ibid.; H. R. 1888 Hearing 8. While the fetus is stuck in this position, dangling partly out of the woman’s body, and just a few inches from a completed birth, the physician uses an instrument such as a pair of scissors to tear or perforate the skull. 11 F. Supp. 2d, at 1106; App. 664 (testimony of Dr. Boehm); Joint Hearing on S. 6 and H. R. 929 before the Senate Committee on the Judiciary and the Subcommittee on the Constitution of the House Committee on the Judiciary, 105th Cong., 1st Sess., 45 (1995) (hereinafter S. 6 and H. R. 929 Joint Hearing). The physician will then either crush the skull or will use a vacuum to remove the brain and other intracranial contents from the fetal skull, collapse the fetus’ head, and pull the fetus from the uterus. 11 F. Supp. 2d, at 1106.7
Use of the partial birth abortion procedure achieved prominence as a national issue after it was publicly described by Dr. Martin Haskell, in a paper entitled “Dilation and Extraction for Late Second Trimester Abortion,” at the National Abortion Federation’s September 1992 Risk Management Seminar. In that paper, Dr. Haskell described his version of the procedure as follows:
“With a lower [fetal] extremity in the vagina, the surgeon uses his fingers to deliver the opposite lower *988extremity, then the torso, the shoulders and the upper extremities.
“The skull lodges at the internal cervical os. Usually there is not enough dilation for it to pass through. The fetus is oriented dorsum or spine up.
“At this point, the right-handed surgeon slides the fingers of the left hand along the back of the fetus and ‘hooks’ the shoulders of the fetus with the index and ring fingers (palm down).
“[T]he surgeon takes a pair of blunt curved Metzen-baum scissors in the right hand. He carefully advances the tip, curved down, along the spine and under his middle finger until he feels it contact the base of the skull under the tip of his middle finger.
“[T]he surgeon then forces the scissors into the base of the skull or into the foramen magnum. Having safely entered the skull, he spreads the scissors to enlarge the opening.
“The surgeon removes the scissors and introduces a suction catheter into this hole and evacuates the skull contents. With the catheter still in place, he applies traction to the fetus, removing it completely from the patient.” H. R. 183S Hearing 3, 8-9.
In cases in which the physician inadvertently dilates the woman to too great a degree, the physician will have to hold the fetus inside the woman so that he can perform the procedure. Id., at 80 (statement of Pamela Smith, M. D.) (“In these procedures, one basically relies on cervical entrapment of the head, along with a firm grip, to help keep the baby in place while the practitioner plunges a pair of scissors into the base of the baby’s skull”). See also S. 6 and H. R. 929 Joint Hearing 45 (“I could put dilapan in for four or five days and say I’m doing a D&E procedure and the fetus could just fall out. But that’s not really the point. The point here is you’re attempting to do an abortion.... Not to see how do *989I manipulate the situation so that I get a live birth instead”) (quoting Dr. Haskell).
II
Nebraska, along with 29 other States, has attempted to ban the partial birth abortion procedure. Although the Nebraska statute purports to prohibit only “partial birth abortion,” a phrase which is commonly used, as I mentioned, to refer to the breech extraction version of intact D&E, the majority concludes that this statute could also be read in some future case to prohibit ordinary D&E, the first procedure described above. According to the majority, such an application would pose a substantial obstacle to some women seeking abortions and, therefore, the statute is unconstitutional. The majority errs with its very first step. I think it is clear that the Nebraska statute does not prohibit the D&E procedure. The Nebraska partial birth abortion statute at issue in this ease reads as follows:
“No partial-birth abortion shall be performed in this state, unless such procedure is necessary to save the life of the mother whose life is endangered by a physical disorder, physical illness, or physical injury, including a life-endangering physical condition caused by or arising from the pregnancy itself.” Neb. Rev. Stat. Ann. §28-328(1) (Supp. 1999).
“Partial birth abortion” is defined in the statute as
“an abortion procedure in which the person performing the abortion partially delivers vaginally a living unborn child before killing the unborn child and completing the delivery. For purposes of this subdivision, the term partially delivers vaginally a living unborn child before killing the unborn child means deliberately and intentionally delivering into the vagina a living unborn child, or a substantial portion thereof, for the purpose of performing a procedure that the person performing such *990procedure knows will kill the unborn child and does kill the unborn child.” §28-326(9).
A
Starting with the statutory definition of "partial birth abortion,” I think it highly doubtful that the statute could be applied to ordinary D&E. First, the Nebraska statute applies only if the physician "partially delivers vaginally a living unborn child,” which phrase is defined to mean “deliberately and intentionally delivering into the vagina a living unborn child, or a substantial portion thereof.” §28-326(9) (emphases added). When read in context, the term "partially delivers” cannot be fairly interpreted to include removing pieces of an unborn child from the uterus one at a time.
The word “deliver,” particularly delivery of an “unborn child,” refers to the process of “assist[ing] in giving birth,” which suggests removing an intact unborn child from the womb, rather than pieces of a child. See Webster's Ninth New Collegiate Dictionary 336 (1991) (defining “deliver” as “to assist in giving birth; to aid in the birth of”); Stedman’s Medical Dictionary 409 (26th ed. 1996) (“To assist a woman in childbirth”). Without question, one does not “deliver” a child when one removes the child from the uterus piece by piece, as in a D&E. Rather, in the words of respondent and his experts, one “remove[s]” or “dismember[s]” the child in a D&E. App. 45, 55 (testimony of Dr. Carhart) (referring to the act of removing the fetus in a D&E); id., at 150 (testimony of Dr. Hodgson) (same); id., at 267 (testimony of Dr. Stubblefield) (physician “dismember[s]” the fetus). See also H. R. 1833 Hearing 3, 8 (Dr. Haskell describing “delivery” of part of the fetus during a D&X). The majority cites sources using the terms “deliver” and “delivery” to refer to removal of the fetus and the placenta during birth. But these sources also presume an intact fetus, rather than dismembered fetal parts. See Obstetrics: Normal & Problem Pregnancies 388 (S. Gabbe, J. Niebyl, & J. Simpson eds., 3d *991ed. 1996) (“After delivery [of infant and placenta], the placenta, cord, and membranes should be examined”); 4 Oxford English Dictionary 421, 422 (2d ed. 1989) (“To disburden (a woman) of the foetus, to bring to childbirth”); B. Maloy, Medical Dictionary for Lawyers 221 (2d ed. 1989) (“To aid in the process of childbirth; to bring forth; to deliver the fetus, placenta”). The majority has pointed to no source in which “delivery” is used to refer to removal of first a fetal arm, then a leg, then the torso, etc. In fact, even the majority describes the D&E procedure without using the word “deliver” to refer to the removal of fetal tissue from the uterus. See ante, at 939 (“pulling a ‘substantial portion’ of -a still living fetus” (emphasis added)); ibid, (“portion of a living fetus has been pulled into the vagina” (emphasis added)). No one, including the majority, understands the act of pulling off a part of a fetus to be a “delivery.”
To make the statute’s meaning even more clear, the statute applies only if the physician “partially delivers vaginally a living unborn child before killing the unborn child and completing the delivery.” The statute defines this phrase to mean that the physician must complete the delivery “for the purpose o/performing a procedure” that will kill the unborn child. It is clear from these phrases that the procedure that kills the fetus must be subsequent to, and therefore separate from, the “partia[l] deliver[y]” or the “deliver[y] into the vagina” of “a living unborn child or substantial portion thereof.” In other words, even if one assumes, arguendo, that dismemberment — the act of grasping a fetal arm or leg and pulling until it comes off, leaving the remaining part of the fetal body still in the uterus — is a kind of “delivery,” it does not take place “before” the death-causing procedure or “for the purpose of performing” the death-causing procedure; it is the death-causing procedure. Under the majority’s view, D&E is covered by the statute because when the doctor pulls on a fetal foot until it tears off he has “delivered” a substantial portion of the unborn child and has performed *992a procedure known to cause death. But, significantly, the physician has not “delivered” the child before performing the death-causing procedure or “for the purpose of” performing the death-causing procedure; the dismemberment “delivery” is itself the act that causes the fetus’ death.8
Moreover, even if removal of a fetal foot or arm from the uterus incidental to severing it from the rest of the fetal body could amount to delivery before, or for the purpose of, performing a death-causing procedure, the delivery would not be of an “unborn child, or a substantial portion thereof.” And even supposing that a fetal foot or arm could conceivably be a “substantial portion” of an unborn child, both the common understanding of “partial birth abortion” and the principle that statutes will be interpreted to avoid constitutional difficulties would require one to read “substantial” otherwise. See infra, at 996-997.
B
Although I think that the text of §28-826(9) forecloses any application of the Nebraska statute to the D&E procedure, even if there were any ambiguity, the ambiguity would be conclusively resolved by reading the definition in light of the fact that the Nebraska statute, by its own terms, applies only to “partial birth abortion,” §28-328(1). By ordinary rules of statutory interpretation, we should resolve any ambiguity in the specific statutory definition to comport with the common understanding of “partial birth abortion,” for that term itself, no less than the specific definition, is part of the stat*993ute. United States v. Morton, 467 U. S. 822, 828 (1984) (“We do not. . . construe statutory phrases in isolation; we read statutes as a whole”).9
“Partial birth abortion” is a term that has been used by a majority of state legislatures, the United States Congress, medical journals, physicians, reporters, even judges, and has never, as far as I am aware, been used to refer to the D&E procedure. The number of instances in which “partial birth abortion” has been equated with the breech extraction form of intact D&E (otherwise known as “D&X”)10 and explicitly contrasted with D&E, are numerous. I will limit myself to just a few examples.
First, numerous medical authorities have equated “partial birth abortion” with D&X. The AMA has done so and has recognized that the procedure is “different from other destructive abortion techniques because the fetus ... is killed outside of the womb.” AMA Board of Trustees Factsheet on H. R. 1122 (June 1997), in App. to Brief for Association of American Physicians and Surgeons et al. as Amici Curiae 1. Medical literature has also equated “partial birth abortion” with D&X as distinguished from D&E. See Gynecologic, Obstetric, and Related Surgery, at 1043; Sprang & Neerhof, Rationale for Banning Abortions Late in Pregnancy, 280 JAMA 744 (Aug. 26,1998); Bopp & Cook, Partial Birth Abortion: The Final Frontier of Abortion Jurisprudence, 14 Issues in Law and Medicine 3 (1998). Physicians have equated “partial birth abortion” with D&X. See Planned Parenthood v. Doyle, 44 F. Supp. 2d 975, 999 (WD Wis. 1999) (citing testimony); Richmond Medical Center for Women v. Gil*994more, 55 F. Supp. 2d 441, 455 (ED Va. 1999) (citing testimony). Even respondent’s expert, Dr. Phillip Stubblefield, acknowledged that breech extraction intact D&E is referred to in the lay press as “partial birth abortion.” App. 271.
Second, the lower courts have repeatedly acknowledged that “partial birth abortion” is commonly understood to mean D&X. See Little Rock Family Planning Services v. Jegley, 192 F. 3d 794, 795 (CA8 1999) (“The term 'partial-birth abortion,’ ... is commonly understood to refer to a particular procedure also known as intact dilation and extraction”); Planned Parenthood of Greater Iowa, Inc. v. Miller, 195 F. 3d 386, 387 (CA8 1999) (“The [Iowa] Act prohibits 'partial-birth abortion,’ a term commonly understood to refer to a procedure called a dilation and extraction (D&X)”). The District Court in this case noted that “[p]artial-birth abortions” are “known medically as intact dilation and extraction or D&X.” 11 F. Supp. 2d, at 1121, n. 26. Even the majority notes that “partial birth abortion” is a term “ordinarily associated with the D&X procedure.” Ante, at 942.
Third, the term “partial birth abortion” has been used in state legislation on 28 occasions and by Congress twice. The term “partial birth abortion” was adopted by Congress in both 1995 and 1997 in two separate pieces of legislation prohibiting the procedure.11 In considering the legislation, *995Congress conducted numerous hearings and debates on the issue, which repeatedly described “partial birth abortion” as a procedure distinct from D&E. The Congressional Record contained numerous references to Dr. Haskell’s procedure. See, e. g., H. R. 1833 Hearing 3,17, 52, 77; S. 6 and H. R. 929 Joint Hearing 45. Since that time, debates have taken place in state legislatures across the country, 30 of which have voted to prohibit the procedure. With only two exceptions, the legislatures that voted to ban the procedure referred to it as “partial birth abortion.”12 These debates also referred to Dr. Haskell’s procedure as D&X. Both the evidence before the legislators and the legislators themselves equated “partial birth abortion” with D&X. The fact that 28 States adopted legislation banning “partial birth abortion,” defined it in a way similar or identical to Nebraska’s definition,13 and, *996in doing so, repeatedly referred to the breech extraction form of intact D&E and repeatedly distinguished it from ordinary D&E, makes it inconceivable that the term “partial birth abortion” could reasonably be interpreted to mean
Were there any doubt remaining whether the statute could apply to a D&E procedure, that doubt is no ground for invalidating the statute. Rather," we are bound to first consider whether a construction of the statute is fairly possible that would avoid the constitutional question. Erznozwik v. Jacksonville, 422 U. S. 205, 216 (1975) (“[A] state statute should not be deemed facially invalid unless it is not readily subject to a narrowing construction by the state courts”); Frisby v. Schultz, 487 U. S. 474, 482 (1988) (“The precise scope of the ban is not further described within the text of the ordinance, but in our view the ordinance is readily subject to a narrowing construction that avoids constitutional difficulties”). This principle is, as Justice O’Connor has said, so “well-established” that failure to apply is “plain error.” Id., at 48S. Although our interpretation of a Nebraska law is of course not binding on Nebraska courts, it is clear, as Erznoz-nik and Frisby demonstrate, that, absent a conflicting interpretation by Nebraska (and there is none here), we should, if the text permits, adopt such a construction.
*997The majority contends that application of the Nebraska statute to D&E would pose constitutional difficulties because it would eliminate the most common form of second-trimester abortions. To the extent that the majority’s contention is true, there is no doubt that the Nebraska statute is susceptible of a narrowing construction by Nebraska courts that would preserve a physicians’ ability to perform D&E. See State v. Carpenter, 250 Neb. 427, 434, 551 N. W. 2d 518, 524 (1996) (“A penal statute must be construed so as to meet constitutional requirements if such can reasonably be done”). For example, the statute requires that the physician "deliberately and intentionally delive[r] into the vagina a living unborn child, or a substantial portion thereof,” before performing a death-causing procedure. The term “substantial portion” is susceptible to a narrowing construction that would exclude the D&E procedure. One definition of the word “substantial” is “being largely but not wholly that which is specified.” Webster’s Ninth New Collegiate Dictionary, at 1176. See Pierce v. Underwood, 487 U. S. 552, 564 (1988) (describing different meanings of the term “substantial”). In other words, “substantial” can mean “almost all” of the thing denominated. If nothing else, a court could construe the statute to require that the fetus be “largely, but not wholly,” delivered out of the uterus before the physician performs a procedure that he knows will kill the unborn child. Or, as I have discussed, a court could (and should) construe “for the purpose of performing a procedure” to mean “for the purpose of performing a separate procedure.”
HH J-H I — I
The majority and Justice O’Connor reject the plain language of the statutory definition, refuse to read that definition in light of the statutory reference to “partial birth abortion,” and ignore the doctrine of constitutional avoidance. In so doing, they offer scant statutory analysis of their own. See ante, at 938-940 (majority opinion); cf. ante, at 940-945 *998(majority opinion); ante, at 948-949 (O’Connor, J., concurring). In their brief analyses, the majority and Justice O’Connor disregard all of the statutory language except for the final definitional sentence, thereby violating the fundamental canon of construction that statutes are to be read as a whole. United States v. Morton, 467 U. S., at 828 (“We do not... construe statutory phrases in isolation; we read statutes as a whole. Thus, the words [in question] must be read in light of the immediately following phrase”) (footnote omitted)); United States v. Heirs of Boisdoré, 8 How. 113, 122 (1849) (“In expounding a statute, we must not be guided by a single sentence or member of a sentence, but look to the provisions of the whole law, and to its object and policy”); Gustafson v. Alloyd Co., 513 U. S. 561, 575 (1995) (“[A] word is known by the company it keeps”).14 In lieu of analyzing the statute as a whole, the majority and Justice O’Connor *999offer five principal arguments for their interpretation of the statute. I will address them in turn.
First, the majority appears to accept, if only obliquely, an argument made by respondent: If the term “partial birth abortion” refers to only the breech extraction form of intact D&E, or D&X, the Nebraska Legislature should have used the medical nomenclature. See ante, at 943 (noting that the Nebraska Legislature rejected an amendment that would replace “partial birth abortion” with “dilation and extraction”); Brief for Respondent 4-5, 24.
There is, of course, no requirement that a legislature use terminology accepted by the medical community. A legislature could, no doubt, draft a statute using the term “heart attack” even if the medical community preferred “myocardial infarction.” Legislatures, in fact, sometimes use medical terms in ways that conflict with their clinical definitions, see, e. g., Barber v. Director, 48 F. 3d 899, 901 (CA4 1995) (noting that the medical definition of “pneumoconiosis” is only a subset of the afflictions that fall within the definition of “pneu-moconiosis” in the Black Lung Act), a practice that is unremarkable so long as the legal term is adequately defined. We have never, until today, suggested that legislature may only use words accepted by every individual physician. Rather, “we have traditionally left to legislators the task of defining terms of a medical nature that have legal significance.” Kansas v. Hendricks, 521 U. S. 346, 359 (1997). And we have noted that “[o]ften, those definitions do not fit precisely with the definitions employed by the medical community.” Ibid.
Further, it is simply not true that the many legislatures, including Nebraska’s, that prohibited “partial birth abortion” chose to use a term known only in the vernacular in place of a term with an accepted clinical meaning. When the Partial-Birth Abortion Ban Act of 1995 was introduced in Congress, the term “dilation and extraction” did not appear in any medical dictionary. See, e. g., Borland’s Illustrated *1000Medical Dictionary 470 (28th ed. 1994); Stedman’s Medical Dictionary, at 485; Miller-Keane Encyclopedia & Dictionary of Medicine, Nursing, & Allied Health 460 (6th ed. 1997); The Sloane-Dorland Annotated Medical-Legal Dictionary 204 (1987); I. Dox, J. Melloni, & G. Eisher, The HarperCollins Illustrated Medical Dictionary 131 (1993). The term did not appear in descriptions of abortion methods in leading medical textbooks. See, e. g., G. Cunningham et al., Williams Obstetrics 579-605 (20th ed. 1997); Obstetrics: Normal & Problem Pregnancies, at 1249-1279; W. Hern, Abortion Practice (1990). Abortion reference books also omitted any reference to the term. See, e. g., Modern Methods of Inducing Abortion (D. Baird, D. Grimes, & P. Van Look eds. 1995); E. Glick, Surgical Abortion (1998).15
Not only did D&X have no medical meaning at the time, but the term is ambiguous on its face. “Dilation and extraction” would, on its face, accurately describe any procedure in which the woman is “dilated” and the fetus “extracted,” including D&E. See supra, at 984-985. In contrast, “partial birth abortion” has the advantage of faithfully describing the procedure the legislature meant to address because the fact that a fetus is “partially born” during the procedure is indisputable. The term “partial birth abortion” is completely accurate and descriptive, which is perhaps the reason why the majority finds it objectionable. Only a desire to find fault at any cost could explain the Court’s willingness to penalize the Nebraska Legislature for failing to replace a *1001descriptive term with a vague one. There is, therefore, nothing to the majority’s argument that the Nebraska Legislature is at fault for declining to use the term “dilation and extraction.”16
Second, the majority faults the Nebraska Legislature for failing to “track the medical differences between D&E and D&X” and for failing to “suggest that its application turns on whether a portion of the fetus’ body is drawn into the vagina as part of a process to extract an intact fetus after collapsing the head as opposed to a process that would dismember the fetus.” Ante, at 939. I have already explained why the Nebraska statute reflects the medical differences between D&X and D&E. To the extent the majority means that the Nebraska Legislature should have “tracked the medical differences” by adopting one of the informal definitions of D&X, this argument is without merit; none of these definitions would have been effective to accomplish the State’s purpose of preventing abortions of partially born fetuses. Take, for example, ACOG’s informal definition of the term “intact D&X.” According to ACOG, an “intact D&X” consists of the following four steps: (1) deliberate dilation of *1002the cervix, usually over a sequence of days; (2) instrumental conversion of the fetus to a footling breach; (8) breech extraction of the body excepting the head; and (4) partial evacuation of the intracranial contents of a living fetus to effect vaginal delivery of a dead but otherwise intact fetus. App. 599-600 (ACOG Executive Board, Statement on Intact Dilation and Extraction (Jan. 12,1997)). ACOG emphasizes that “unless all four elements are present in sequence, the procedure is not an intact D&X.” Id., at 600. Had Nebraska adopted a statute prohibiting “intact D&X,” and defined it along the lines of the ACOG definition, physicians attempting to perform abortions on partially born fetuses could have easily evaded the statute. Any doctor wishing to perform a partial birth abortion procedure could simply avoid liability under such a statute by performing the procedure, as respondent does, only when the fetus is presented feet first, thereby avoiding the necessity of “conversion of the fetus to a footling breech.” Id., at 599. Or, a doctor could convert the fetus without instruments. Or, the doctor could cause the fetus’ death before “partial evacuation of the intracranial contents,” id., at 600, by plunging scissors into the fetus’ heart, for example. A doctor could even attempt to evade the statute by chopping off two fetal toes prior to completing delivery, preventing the State from arguing that the fetus was “otherwise intact.” Presumably, however, Nebraska, and the many other legislative bodies that adopted partial birth abortion bans, were not concerned with whether death was inflicted by injury to the brain or the heart, whether the fetus was converted with or without instruments, or whether the fetus died with its toes attached. These legislative bodies were, I presume, concerned with whether the child was partially born before the physician caused its death. The legislatures’ evident concern was with permitting a procedure that resembles infanticide and threatens to dehumanize the fetus. They, therefore, presumably declined to adopt a *1003ban only on “intact D&X,” as defined by ACOG, because it would have been ineffective to that purpose. Again, the majority is faulting Nebraska for a legitimate legislative calculation.
Third, the majority and Justice O’Connor argue that this Court generally defers to lower federal courts’ interpretations of state law. Ante, at 940 (majority opinion); ante, at 949 (O’Connor, J., concurring). However, a decision drafted by Justice O’Connor, which she inexplicably fails to discuss, Frisby v. Schultz, 487 U. S. 474 (1988), makes clear why deference is inappropriate here. As Justice O’Con-nor explained in that case:
“[Wjhile we ordinarily defer to lower court constructions of state statutes, we do not invariably do so. We are particularly reluctant to defer when the lower courts have fallen into plain error, which is precisely the situation presented here. To the extent they endorsed a broad reading of the ordinance, the lower courts ran afoul of the well-established principle that statutes will be interpreted to avoid constitutional difficulties.” Id., at 483 (citations omitted).
Frisby, then, identifies exactly why the lower courts’ opinions here are not entitled to deference: The lower courts failed to identify the narrower construction that, consistent with the text, would avoid any constitutional difficulties.
Fourth, the majority speculates that some Nebraska prosecutor may attempt to stretch the statute to apply it to D&E. But a state statute is not unconstitutional on its face merely because we can imagine an aggressive prosecutor who would attempt an overly aggressive application of the statute. We have noted that “‘[wjords inevitably contain germs of uncertainty.’ ” Broadrick v. Oklahoma, 413 U. S. 601, 608 (1973). We do not give statutes the broadest definition imaginable. Rather, we ask whether “the ordinary *1004person exercising ordinary common sense can sufficiently understand and comply with [the statute].” Ibid, [quoting Civil Service Comm’n v. Letter Carriers, 413 U. S. 548, 579 (1973)). While a creative legal mind might be able to stretch the plain language of the Nebraska statute to apply to D&E, “citizens who desire to obey the statute will have no difficulty in understanding it.” Colten v. Kentucky, 407 U. S. 104,110 (1972) (internal quotation marks omitted).
Finally, the majority discusses at some length the reasons it will not defer to the interpretation of the statute proffered by the Nebraska Attorney General, despite the Attorney General’s repeated representations to this Court that his State will not apply the partial birth abortion statute to D&E. See Brief for Petitioners 11-13; Tr. of Oral Arg. 10-11. The fact that the Court declines to defer to the interpretation of the Attorney General is not, however, a reason to give the statute a contrary representation. Even without according the Attorney General’s view any particular respect, we should agree with his interpretation because it is undoubtedly the correct one. Moreover, Justice O’Connor has noted that the Court should adopt a narrow interpretation of a state statute when it is supported by the principle that statutes will be interpreted to avoid constitutional difficulties as well as by “the representations of counsel... at oral argument.” Frisby v. Schultz, supra, at 483. Such an approach is particularly appropriate in this case because, as the majority notes, Nebraska courts accord the Nebraska Attorney General’s interpretations of state statutes “substantial weight.” See State v. Coffman, 213 Neb. 560, 561, 330 N. W. 2d 727, 728 (1983). Therefore, any renegade prosecutor bringing criminal charges against a physician for performing a D&E would find himself confronted with a contrary interpretation of the statute by the Nebraska Attorney General, and, I assume, a judge who both possessed common *1005sense and was aware of the rule of lenity. See State v. White, 254 Neb. 566, 575, 577 N. W. 2d 741,747 (1998).17
rH <!
Having resolved that Nebraskas partial birth abortion statute permits doctors to perform D&E abortions, the question remains whether a State can constitutionally prohibit the partial birth abortion procedure without a health exception. Although the majority and Justice O’Connor purport to rely on the standard articulated in the Casey plurality in concluding that a State may not, they in fact disregard it entirely.
A
Though Justices O’Connor, Kennedy, and Souter declined in Casey, on the ground of stare decisis, to reconsider whether abortion enjoys any constitutional protection, 505 U. S., at 844-846,854-869 (majority opinion); id., at 871 (plurality opinion), Casey professed to be, in part, a repudiation of Roe and its progeny. The Casey plurality expressly noted that prior case law had undervalued the State’s interest in potential life, 505 U. S., at 875-876, and had invalidated regulations of abortion that “in no real sense deprived women of the ultimate decision,” id., at 875. See id., at 871 (“Roe v. Wade speaks with clarity in establishing ... the State’s ‘important and legitimate interest in potential life.’ That por*1006tion of the decision in Roe has been given too little acknowledgment” (citation omitted)). The plurality repeatedly recognized the States’ weighty interest in this area. See id., at 877 (“State ... may express profound respect for the life of the unborn”); id., at 878 (“the State’s profound interest in potential life”); id., at 850 (majority opinion) (“profound moral and spiritual implications of terminating a pregnancy, even in its earliest stage”). And, the plurality expressed repeatedly the States’ legitimate role in regulating abortion procedures. See id., at 876 (“The very notion that the State has a substantial interest in potential life leads to the conclusion that not all regulations must be deemed unwarranted”); id., at 875 (“Not all governmental intrusion [with abortion] is of necessity unwarranted”). According to the plurality: “The fact that a law which serves a valid purpose, one not designed to strike at the right itself, has the incidental effect of making it more difficult or more expensive to procure an abortion cannot be enough to invalidate it.” Id., at 874.
The Casey plurality therefore adopted the standard: “Only where state regulation imposes an undue burden on a woman’s ability to make this decision does the power of the State reach into the heart of the liberty protected by the Due Process Clause.” Ibid. A regulation imposes an “undue burden” only if it “has the effect of placing a substantial obstacle in the path of a woman’s choice.” Id., at 877.
B
There is no question that the State of Nebraska has a valid interest — one not designed to strike at the right itself — in prohibiting partial birth abortion. Casey itself noted that States may “express profound respect for the life of the unborn.” Ibid. States may, without a doubt, express this profound respect by prohibiting a procedure that approaches infanticide, and thereby dehumanizes the fetus and trivializes human life. The AMA has recognized that this procedure is “ethically different from other destructive abortion *1007techniques because the fetus, normally twenty weeks or longer in gestation, is killed outside the womb. The 'partial birth’ gives the fetus an autonomy which separates it from the right of the woman to choose treatments for her own body.” AMA Board of Trustees Factsheet on H. R. 1122 (June 1997), in App. to Brief for Association of American Physicians and Surgeons et al. as Amici Curiae 1. Thirty States have concurred with this view.
Although the description of this procedure set forth above should be sufficient to demonstrate the resemblance between the partial birth abortion procedure and infanticide, the testimony of one nurse who observed a partial birth abortion procedure makes the point even more vividly:
"The baby’s little fingers were clasping and unclasping, and his little feet were kicking. Then the doctor stuck the scissors in the back of his head, and the baby’s arms jerked out, like a startle reaction, like a flinch, like a baby does when he thinks he is going to fall.
“The doctor opened up the scissors, stuck a high-powered suction tube into the opening, and sucked the baby’s brains out. Now the baby went completely limp.” H. R. 1833 Hearing 18 (statement of Brenda Pratt Shafer).
The question whether States have a legitimate interest in banning the procedure does not require additional authority. See ante, at 961-964 (Kennedy, J., dissenting).18 In a civi*1008lized society, the answer is too obvious, and the contrary arguments too offensive, to merit further discussion. But see ante, at 946-947 (Stevens, J., concurring) (arguing that the decision of BO States to ban the partial birth abortion procedure was “simply irrational” because other forms of abortion were “equally gruesome”); ante, at 961-952 (Ginsburg, J., concurring) (similar).19
*1009C
The next question, therefore, is whether the Nebraska statute is unconstitutional because it does not contain an exception that would allow use of the procedure whenever “ ‘ “necessary, in appropriate medical judgment, for the preservation of the . . . health of the mother.” ’ ” Ante, at 930 (majority opinion) (quoting Casey, 505 U. S., at 879, in turn quoting Roe, 410 U. S., at 164-165) (emphasis deleted). According to the majority, such a health exception is required here because there is a “division of opinion among some medical experts over whether D&X is generally safer [than D&E], and an absence of controlled medical studies that would help answer these medical questions.” Ante, at 936-937. In other words, unless a State can conclusively establish that an abortion procedure is no safer than other procedures, the State cannot regulate that procedure without including a health exception. Justice O’Connor agrees. Ante, at 947-948 (concurring opinion). The rule set forth by the majority and Justice O’Connor dramatically expands on our prior abortion cases and threatens to undo any state regulation of abortion procedures.
The majority and Justice O’Connor suggest that their rule is dictated by a straightforward application of Roe and Casey. Ante, at 929-930 (majority opinion); ante, at 947-948 (O’Connor, J., concurring). But that is simply not true. In Roe and Casey, the Court stated that the State may “regulate, and even proscribe, abortion except where it is necessary, in appropriate medical judgment, for the preservation of the life or health of the mother.” Roe, supra, at 165; Casey, 505 U. S., at 879. Casey said that a health exception must be available if “continuing her pregnancy would constitute a threat” to the woman. Id., at 880 (majority opinion) (emphasis added). Under these eases, if a State seeks to prohibit abortion, even if only temporarily or under particular circumstances, as Casey says that it may, id., at 879 (plu*1010rality opinion), the State must make an exception for cases in which the life or health of the mother is endangered by continuing the pregnancy. These cases addressed only the situation in which a woman must obtain an abortion because of some threat to her health from continued pregnancy. But Roe and Casey say nothing at all about eases in which a physician considers one prohibited method of abortion to be preferable to permissible methods. Today's majority and Justice O'Connor twist Roe and Casey to apply to the situation in which a woman desires — for whatever reason — an abortion and wishes to obtain the abortion by some particular method. See ante, at 929-931 (majority opinion); ante, at 947-948 (concurring opinion). In other words, the majority and Justice O'Connor fail to distinguish between cases in which health concerns require a woman to obtain an abortion and cases in which health concerns cause a woman who desires an abortion (for whatever reason) to prefer one method over another.
It is clear that the Court's understanding of when a health exception is required is not mandated by our prior cases. In fact, we have, post-Casey, approved regulations of methods of conducting abortion despite the lack of a health exception. Mazurek v. Armstrong, 520 U. S. 968, 971 (1997) (per cu-riam) (reversing Court of Appeals holding that plaintiffs challenging requirement that only physicians perform abortions had a “'fair chance of success’”); id., at 979 (Stevens, J., dissenting) (arguing that the regulation was designed to make abortion more difficult). And one can think of vast bodies of law regulating abortion that are valid, one would hope, despite the lack of health exceptions. For example, physicians are presumably prohibited from using abortifa-cients that have not been approved by the Food and Drug Administration even if some physicians reasonably believe *1011that these abortifaeients would be safer for women than existing abortifaeients.20
The majority effectively concedes that Casey provides no support for its broad health exception rule by relying on live-Casey authority, see ante, at 931, including a case that was specifically disapproved of in Casey for giving too little weight to the State’s interest in fetal life. See Casey, supra, at 869,882 (overruling the parts of Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S. 747 (1986), that were “inconsistent with Roe’s statement that the State has a legitimate interest in promoting the life or potential life of the unborn,” 506 U. S., at 870); id., at 893 (majority opinion) (relying on Thornburgh, supra, at 783 (Burger, C. J., dissenting), for the proposition that the Court was expanding on Roe in that case). Indeed, Justice O’Connor, who joins the Court’s opinion, was on the Court for Thornburgh and was in dissent, arguing that, under the undue-burden standard, the statute at issue was constitutional. See 476 U. S., at 828-832 (arguing that the challenged state statute was not “unduly burdensome”). The majority’s resort to this ease proves my point that the holding today assumes that the standard set forth in the Casey joint opinion is no longer governing.
And even if I were to assume that the pre-Casey standards govern, the cases cited by the majority provide no support for the proposition that the partial birth abortion ban must *1012include a health exception because some doctors believe that partial birth abortion is safer. In Thornburgh, Danforth, and Doe, the Court addressed health exceptions for cases in which continued pregnancy would pose a risk to the woman. Thornburgh, supra, at 770; Planned Parenthood of Central Mo. v. Danforth, 428 U. S. 52 (1976); Doe v. Bolton, 410 U. S., at 197.. And in Colautti v. Franklin, 439 U. S. 379 (1979), the Court explicitly declined to address whether a State can constitutionally require a tradeoff between the woman’s health and that of the fetus. The broad rule articulated by the majority and by Justice O’Connor are unprecedented expansions of this Court’s already expansive pre-Casey jurisprudence.
As if this state of affairs were not bad enough, the majority expands the health exception rule articulated in Casey in one additional and equally pernicious way. Although Roe and Casey mandated a health exception for cases in which abortion is “necessary” for a woman’s health, the majority concludes that a procedure is “necessary” if it has any comparative health benefits. Ante, at 937. In other words, according to the majority, so long as a doctor can point to support in the profession for his (or the woman’s) preferred procedure, it is “necessary” and the physician is entitled to perform it. Ibid. See also ante, at 952 (Ginsburg, J., concurring) (arguing that a State cannot constitutionally “sto[p] a woman from choosing the procedure her doctor ‘reasonably believes’ ” is in her best interest). But such a health exception requirement eviscerates Casey’s undue-burden standard and imposes unfettered abortion on demand. The exception entirely swallows the rule. In effect, no regulation of abortion procedures is permitted because there will always be some support for a procedure and there will always be some doctors who conclude that the procedure is preferable. If Nebraska reenacts its partial birth abortion ban with a health exception, the State will not be able to prevent physicians like Dr. Carhart from using partial birth abortion as a routine abortion procedure. This Court has now expressed *1013its own conclusion that there is “highly plausible” support for the view that partial birth abortion is safer, which, in the majority’s view, means that the procedure is therefore “necessary.” Ante, at 937. Any doctor who wishes to perform such a procedure under the new statute will be able to do so with impunity. Therefore, Justice O’Connor’s assurance that the constitutional failings of Nebraska’s statute can be easily fixed, ante, at 950-951, is illusory. The majority’s insistence on a health exception is a fig leaf barely covering its hostility to any abortion regulation by the States — a hostility that Casey purported to reject.21
D
The majority assiduously avoids addressing the actual standard articulated in Casey — whether prohibiting partial birth abortion without a health exception poses a substantial obstacle to obtaining an abortion. 505 U. S., at 877. And for good reason: Such an obstacle does not exist. There are two essential reasons why the Court cannot identify a substantial obstacle. First, the Court cannot identify any real, much less substantial, barrier to any woman’s ability to obtain an abortion. And second, the Court cannot demonstrate that any such obstacle would affect a sufficient number of women to justify invalidating the statute on its face.
1
The Casey joint opinion makes clear that the Court should not strike down state regulations of abortion based on the *1014fact that some women might face a marginally higher health risk from the regulation. In Casey, the Court upheld a 24-hour waiting period even though the Court credited evidence that for some women the delay would, in practice, be much longer than 24 hours, and even though it was undisputed that any delay in obtaining an abortion would impose additional health risks. Id., at 887; id., at 937 (Blackmun, J., concurring in part, concurring in judgment in part, and dissenting in part) (“The District Court found that the mandatory 24-hour delay could lead to delays in excess of 24 hours, thus increasing health risks”). Although some women would be able to avoid the waiting period because of a “medical emergency,” the medical emergency exception in the statute was limited to those women for whom delay would create “serious risk of substantial and irreversible impairment of a major bodily function.” Id., at 902 (appendix to joint opinion) (internal quotation marks omitted). Without question, there were women for whom the regulation would impose some additional health risk who would not fall within the medical emergency exception. The Court concluded, despite the certainty of this increased risk, that there was no showing that the burden on any of the women was substantial. Id., at 887.
The only ease in which this Court has overturned a State’s attempt to prohibit a particular form of abortion also demonstrates that a marginal increase in health risks is not sufficient to create an undue burden. In Planned Parenthood of Central Mo. v. Danforth, 428 U. S. 52 (1976), the Court struck down a state regulation because the State had outlawed the method of abortion used in 70% of abortions and because alternative methods were, the Court emphasized, “significantly more dangerous and critical” than the prohibited method. Id., at 76.
Like the Casey 24-hour waiting period, and in contrast to the situation in Danforth, any increased health risk to women imposed by the partial birth abortion ban is minimal *1015at most. Of the 5.5% of abortions that occur after 15 weeks (the time after which a partial birth abortion would be possible), the vast majority are performed with a D&E or induction procedure. And, for any woman with a vertex presentation fetus, the vertex presentation form of intact D&E, which presumably shares some of the health benefits of the partial birth abortion procedure but is not covered by the Nebraska statute, is available. Of the remaining women— that is, those women for whom a partial birth abortion procedure would be considered and who have a breech presentation fetus — there is no showing that any one faces a significant health risk from the partial birth abortion ban. A select committee of ACOG “could identify no circumstances under which this procedure ... would be the only option to save the life or preserve the health of the woman.” App. 600 (ACOG Executive Board, Statement on Intact Dilation and Extraction (Jan. 12, 1997))* See also Hope Clinic v. Ryan, 195 F. 8d 857, 872 (CA7 1999) (en banc) (“ ‘There does not appear to be any identified situation in which intact D&X is the only appropriate procedure to induce abortion’ ” (quoting Late Term Pregnancy Techniques, AMA Policy H-5.982 W. D. Wis. 1999)); Planned Parenthood of Wis. v. Doyle, 44 F. Supp. 2d, at 980 (citing testimony of Dr. Haskell that “the D&X procedure is never medically neeessary to... preserve the health of a woman”), vacated, 195 F. 3d 857 (CA7 1999). And, an ad hoe coalition of doctors, including former Surgeon General Koop, concluded that there are-no medical conditions that require use of the partial birth abortion procedure to preserve the mother’s health. See App. 719.
In fact, there was evidence before the Nebraska Legislature that partial birth abortion increases health risks relative to other procedures. During floor debates, a proponent of the Nebraska legislation read from and cited several articles by physicians concluding that-partial birth abortion procedures are risky. App. in Nos. 98-8245, 98-3300 (GA8), *1016p. 812. One doctor testifying before a committee of the Nebraska Legislature stated that partial birth abortion involves three “very risky procedures”: dilation of the cervix, using instruments blindly, and conversion of the fetus. App. 721 (quoting testimony of Paul Hays, M. D.).22
There was also evidence before Congress that partial birth abortion “does not meet medical standards set by ACOG nor has it been adequately proven to be safe nor efficacious.” H. R. 1838 Hearing 112 (statement of Nancy G. Romer, M. D.); see id., at 110-111.23 The AMA supported the congressional ban on partial birth abortion, concluding that the procedure is “not medically indicated” and “not good medicine.” See 143 Cong. Ree. S4670 (May 19,1997) (reprinting a letter from the AMA to Sen. Santorum). And there was evidence before Congress that there is “certainly no basis upon which to state the claim that [partial birth abortion] is a safer or even a preferred procedure.” Partial Birth Abortion: The Truth, S. 6 and H. R. 929 Joint Hearing 123 (statement of Curtis Cook, M. D.). This same doctor testified that *1017“partial-birth abortion is an unnecessary, unsteady, and potentially dangerous procedure,” and that “safe alternatives are in existence.” Id., at 122.
The majority justifies its result by asserting that a “significant body of medical opinion” supports the view that partial birth abortion may be a safer abortion procedure. Ante, at 937. I find this assertion puzzling. If there is a “significant body of medical opinion” supporting this procedure, no one in the majority has identified it. In fact, it is uncontested that although this procedure has been used since at least 1992, no formal studies have compared partial birth abortion with other procedures. 11F. Supp. 2d, at 1112 (citing testimony of Dr. Stubblefield); id., at 1115 (citing testimony of Dr. Boehm); Epner, Jonas, & Seckinger, Late-term Abortion, 280 JAMA 724 (Aug. 26,1998); Sprang & Neerhof, Rationale for Banning Abortion Late in Pregnancy, 280 JAMA 744 (Aug, 26, 1998). Cf. Kumko Tire Co. v. Carmichael, 526 U. S. 137,149-152 (1999) (observing that the reliability of a scientific technique may turn on whether the technique can be and has been tested; whether it has been subjected to peer review and publication; and whether there is a high rate of error or standards controlling its operation). The majority’s conclusion makes sense only if the undue-burden standard is not whether a “significant body of medical opinion” supports the result, but rather, as Justice Ginsburg candidly admits, whether any doctor could reasonably believe that the partial birth abortion procedure would best protect the woman. Ante, at 952.
Moreover, even if I were to assume credible evidence on both sides of the debate, that fact should resolve the undue-burden question in favor of allowing Nebraska to legislate. Where no one knows whether a regulation of abortion poses any burden at all, the burden surely does not amount to a “substantial obstacle.” Under Casey, in sueh a case we should defer to the legislative judgment. We have said:
*1018“[I]t is precisely where such disagreement exists that legislatures have been afforded the widest latitude in drafting such statutes. ... [W]hen a legislature undertakes to act in areas fraught with medical and scientific uncertainties, legislative options must be especially broad_” Kansas v. Hendricks, 521U. S., at 360, n. 3 (internal quotations marks omitted).
In Justice O’Connor’s words:
“It is . . . difficult to believe that this Court, without the resources available to those bodies entrusted with making legislative choices, believes itself competent to make these inquiries and to revise these standards every time the American College of Obstetricians and Gynecologists (ACOG) or similar group revises its views about what is and what is not appropriate medical procedure in this area.” Akron v. Akron Center for Reproductive Health, Inc., 462 U. S., at 456 (dissenting opinion).
See id., at 456, n. 4 (“Irrespective of the difficulty of the task, legislatures, with their superior factfinding capabilities, are certainly better able to make the necessary judgments than are courts”); Webster v. Reproductive Health Services, 492 U. S., at 519 (plurality opinion) (Court should not sit as an “ex officio medical board with powers to approve or disapprove medical and operative practices and standards throughout the United States (internal quotations marks omitted)); Jones v. United States, 463 U. S. 354, 365, n. 13 (1983) (“The lesson we have drawn is not that government may not act in the face of this [medical] uncertainty, but rather that courts should pay particular deference to reasonable legislative judgments”). The Court today disregards these principles and the clear import of Casey.
2
Even if I were willing to assume that the partial birth method of abortion is safer for some small set of women, such *1019a conclusion would not require invalidating the Act, because this case comes to us on a facial challenge. The only question before us is whether respondent has shown that “'no set of circumstances exists under which the Act would be valid/ ” Ohio v. Akron Center for Reproductive Health, 497 U. S. 502, 514 (1990) (quoting Webster v. Reproductive Health Services, supra, at 524 (O’Connor, J., concurring in part and concurring in judgment)). Courts may not invalidate on its face a state statute regulating abortion “based upon a worst-case analysis that may never occur.” 497 U. S., at 514.
Invalidation of the statute would be improper even assuming that Casey rejected this standard sub silentio (at least so far as abortion cases are concerned) in favor of a so-called “ ‘large fraction’ ” test. See Fargo Women’s Health Organization v. Schafer, 507 U. S. 1013, 1014 (1993) (O’Connor, J., joined by Souter, J., concurring) (arguing that the “no set of circumstances” standard is incompatible with Casey). See also Janklow v. Planned Parenthood, Sioux Falls Clinic, 517 U. S. 1174, 1177-1179 (1996) (Scalia, J., dissenting from denial of certiorari). In Casey, the Court was presented with a facial challenge to, among other provisions, a spousal notice requirement. The question, according to the majority, was whether the spousal notice provision operated as a “substantial obstacle” to the women “whose conduct it affects,” namely, “married women seeking abortions who do not wish to notify their husbands of their intentions and who do not qualify for one of the statutory exceptions to the notice requirement.” 505 U. S., at 895. The Court determined that a “large fraction” of the women in this category were victims of psychological or physical abuse. Ibid. For this subset of women, according to the Court, the provision would pose a substantial obstacle to the ability to obtain an abortion because their husbands could exercise an effective veto over their decision. Id., at 897.
None of the opinions supporting the majority so much as mentions the large fraction standard, undoubtedly because *1020the Nebraska statute easily survives it. I will assume, for the sake of discussion, that the category of women whose conduct Nebraska’s partial birth abortion statute might affect includes any woman who wishes to obtain a safe abortion after 16 weeks’ gestation. I will also assume (although I doubt it is true) that, of these women, every one would be willing to use the partial birth abortion procedure if so advised by her doctor. Indisputably, there is no “large fraction” of these women who would face a substantial obstacle to obtaining a safe abortion because of their inability to use this particular procedure. In fact, it is not clear that any woman would be deprived of a safe abortion by her inability to obtain a partial birth abortion. More medically sophisticated minds than ours have searched and failed to identify a single circumstance (let alone a large fraction) in which partial birth abortion is required. But no matter. The “ad hoe nullification” machine is back at full throttle. See Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S., at 814 (O’Connor, J., dissenting); Madsen v. Women's Health Center, Inc., 512 U. S. 753, 785 (1994) (Scalia, J., concurring in judgment in part and dissenting in part).
•i* H’ ^
We were reassured repeatedly in Casey that not all regulations of abortion are unwarranted and that the States may express profound respect for fetal life. Under Casey, the regulation before us today should easily pass constitutional muster. But the Court’s abortion jurisprudence is a particularly virulent strain of constitutional exegesis. And so today we are told that 30 States are prohibited from banning one rarely used form of abortion that they believe to border on infanticide. It is clear that the Constitution does not compel this result.
I respectfully dissent.
Unless otherwise noted, all subsequent cites of Planned Parenthood of Southeastern Pa. v. Casey, 505 U. S. 833 (1992), are to the joint opinion of O’Connor, Kennedy, and Souter, JJ.
in 1996, the most recent year for which abortion statistics are available from the Centers for Disease Control and Prevention, there were approximately 1,221,585 abortions performed in the United States. Centers for Disease Control and Prevention, Abortion Surveillance — United States, 1996, p. 1 (July 80, 1999). Of these abortions, about 67,000 — 5.5%—were performed in or after the 16th week of gestation, that is, from the middle of the second trimester through the third trimester. Id, at 5. The majority apparently accepts that none of the abortion procedures used for pregnancies in earlier stages of gestation, including “dilation and evacuation” (D&E) as it is practiced between 13 and 15 weeks’ gestation, would be compromised by the statute. See ante, at 938-940 (concluding that the statute could be interpreted to apply to instrumental dismemberment procedures used in a later term D&E). Therefore, only the methods of abortion available to women in this later stage of pregnancy are at issue in this ease.
At 16 weeks’ gestation, the average fetus is approximately six inches long. By 20 weeks’ gestation, the fetus is approximately eight inches long. K Moore & T. Persaud, The Developing Human 112 (6th ed. 1998).
Past the 20th week of gestation, respondent attempts to induce fetal death by injection prior to beginning the procedure in patients. 11 F. Supp. 2d, at 1106; App. 64.
There is apparently no general understanding of which women are appropriate candidates for the procedure. Respondent uses the procedure on women at 16 to 20 weeks’ gestation. 11 F. Supp. 2d, at 1105. The doctor who developed the procedure, Dr. Martin Haskell, indicated that he *987performed the procedure on patients 20 through 24 weeks and on certain patients 25 through 26 weeks. See H. R. 1833 Hearing 36.
There are, in addition, two forms of abortion that are used only rarely: hysterotomy, a procedure resembling a Caesarean section, requires the surgical delivery of the fetus through an incision on the uterine wall, and hysterectomy. 11 F. Supp. 2d, at 1109.
The majority argues that the statute does not explicitly require that the death-causing procedure be separate from the overall abortion procedure. That is beside the point; under the statute the death-causing procedure must be separate from the delivery. Moreover, it is incorrect to state that the statute contemplates only one “procedure.” The statute clearly uses the term “procedure” to refer to both the overall abortion procedure (“partial birth abortion” is “an abortion procedure”) as well as to a component of the overall abortion procedure (“for the purpose of performing a procedure... that will kill the unborn child”).
It is certainly true that an undefined term must be construed in accordance with its ordinary and plain meaning. FDIC v. Meyer, 510 U. S. 471, 476 (1994). But this does not mean that the ordinary and plain meaning of a term is wholly irrelevant when that term is defined.
As noted, see n. 5, swpm, there is no consensus regarding which of these terms is appropriate to describe the procedure. I assume, as the majority does, that the terms are, for purposes here, interchangeable.
Congressional legislation prohibiting the procedure was first introduced in June 1995, with the introduction of the Partial Birth Abortion Ban Act, H. R. 1833. This measure, which was sponsored by 165 individual House Members, passed both Houses by wide margins, 141 Cong. Rec. 35892 (1995); 142 Cong. Rec. 31169 (1996), but was vetoed by President Clinton, see id., at 7467. The House voted to overridé the veto on September 19,1996, see id., at 23851; however, the Senate failed to override by a margin of 13 votes, see id., at 25829. In the next Congress, 181 individual House cosponsors reintroduced the Partial Birth Abortion Ban Act as H. R. 929, which was later replaced in the House with H. R. 1122. See H. R. 1122, 105th Cong., 1st Sess. (1997). The House and Senate again adopted the legislation, as amended, by wide margins. See 143 *995Cong. Rec. H1230 (Mar. 20,1997); id., at S4715 (May 20,1997). President Clinton again vetoed the bill. See id., at H8891 (Oct. 10,1997). Again, the veto override passed in the House and fell short in the Senate. See 144 Cong. Ree. H6213 (July 23,1998); id., at S10564 (Sept. 18,1998).
Consistent with the practice of Dr. Haskell (an Ohio practitioner), Ohio referred to the procedure as “dilation and extraction,” defined as “the termination of a human pregnancy by purposely inserting a suction device into the skull of a fetus to remove the brain.” Ohio Rev. Code Ann. § 2919.15(A) (1997). Missouri refers to the killing of a “partially-born” infant as “infanticide.” Mo. Stat. Ann. §565.300 (Vernon Supp. 2000).
For the most part, these States defined the term “partial birth abortion” using language similar to that in the 1995 proposed congressional legislation, that is “an abortion in which the person performing the abortion partially vaginally delivers a living fetus before killing the fetus and completing the delivery.” See H R. 1833 Hearing 210. See, e. g., Alaska Stat. Ann. §18.16.050 (1998); Ariz. Rev. Stat. Ann. §13-3603.01 (Supp. 1999); Ark. Code Ann. §5-61-202 (1997); Fla. Stat. §390.011 (Supp. 2000); Ill. Comp. Stat., ch. 720, §513/5 (1999); Ind. Code Ann. §16-18-2-267.5 (West Supp. 1999); Mich. Comp. Laws Ann. §333.17016(5)(c) (West Supp. 2000); Miss. Code Ann. §41-41-73(2)(a) (Supp. 1998); S. C. Code Ann. §44-41-85(A)(1) (1999 Cum. Supp.). Other States, including Nebraska, see Neb. Rev. Stat. Ann. §28-326 (Supp. 1999), defined “partial-birth abortion” using language similar to that used in the 1997 proposed congressional *996legislation, which retained the definition of partial birth abortion used in the 1995 bill, that is “an abortion in which the person performing the abortion partially vaginally delivers a living fetus before killing the fetus and completing the delivery,” but further defined that phrase to mean “deliberately and intentionally delivers into the vagina a living fetus, or a substantial portion there of, for the purpose of performing a procedure the physician knows will kill the fetus, and Mils the fetus.” See Partial Birth Abortion Ban Act of 1997, H. R. 1122, 105th Cong., 1st Sess. (1997). See, e.g., Idaho Code §18-613(a) (Supp. 1999); Iowa Code Ann. § 707.8A(l)(e) (Supp. 1999); N. J. Stat. Ann. §2A:65A-6(e) (West Supp. 2000); OMa. Stat. Ann., Tit. 21, §684 (Supp. 2000); R. I. Gen. Laws §23-4.12-1 (Supp. 1999); Tenn. Code Ann. §39-15-209(a)(l) (1997).
The majority argues that its approach is supported by Meese v. Keene, 481 U. S. 465, 487 (1987), in which the Court stated that “the statutory definition of [a] term excludes unstated meanings of that term.” But this case provides no support for the approach adopted by the majority and Justice O’Connor. In Meese, the Court addressed a statute that used the term “political propaganda.” Id., at 470. The Court noted that there were two commonly understood meanings to the term “political propaganda,” id., at 477, and, not surprisingly, chose the definition that was most consistent with the statutory definition, id., at 485. Nowhere did the Court suggest that, because “political propaganda” was defined in the statute, the commonly understood meanings of that term were irrelevant. Indeed, a significant portion of the Court’s opinion was devoted to describing the effect of Congress’use of that term. id., at 477-479, 488-484. So too, Colautti v. Franklin, 489 U. S. 879, 392-393, n. 10 (1979), and 'Western Union Telegraph, Co. v. Lenroot, 323 U. S. 490 (1945), support the proposition that when there are two possible interpretations of a term, and only one comports with the statutory definition, the term should not be read to include the unstated meaning. But here, there is only one possible interpretation of “partial birth abortion” — the majority can cite no authority using that term to describe D&E — and so there is no justification for the majority’s willingness to entirely disregard the statute’s use of that term.
Nor, for that matter, did the terms “intact dilation and extraction” or “intact dilation and evacuation” appear in textbooks or medical dictionaries. See supra, at 999 and this page. In fact, respondent’s preferred term “intact D&E” would compound, rather than remedy, any confesión regarding the statute’s meaning. As is evident from the majority opinion, there is no consensus on what this term means. Compare ante, at 927 (describing “intact D&E” to refer to both breech and vertex presentation procedures), with App. 6 (testimony of Dr. Henshaw) (using “intact D&E” to mean only breech procedure), with id., at 275 (testimony of Dr. Stub-blefield) (using “intact D&E” to refer to delivery of fetus that has died in útero).
The fact that the statutory term “partial birth abortion” may express a political or moral judgment, whereas “dilation and extraction” does not, is irrelevant. It is certainly true that technical terms are frequently empty of normative content. (Of course, the decision to use a technical term can itself be normative. See ante, passim (majority opinion)). But, so long as statutory terms are adequately defined, there is no requirement that Congress or state legislatures draft statutes using morally agnostic terminology. See, e. g., 18 U. S. C. §922(v) (making it unlawful to “manufacture, transfer, or possess a semiautomatic assault weapon”); Kobay-ashi & Olson et al., In re 101 California Street: A Legal and Economic Analysis of Strict Liability for the Manufacture and Sale of “Assault Weapons,” 8 Stan. L. & Pol’y Rev. 41,43 (1997) (“Prior to 1989, the term 'assault weapon’ did not exist in the lexicon of firearms. It is a political term, developed by anti-gun publicists to expand the category of 'assault rifles’ so as to allow an attack on as many additional firearms as possible on the basis of undefined ‘evil’ appearance”). See also Meese, 481 U. S., at 484-485.
The majority relies on Justice Scalia’s observation in Crandon v. United, States, 494 U. S. 152 (1990), that “we have never thought that the interpretation of those charged with prosecuting criminal statutes is entitled to deference.” Id., at 177. But Justice Scalia was commenting on the United States Attorney General's overly broad interpretation of a federal statute, deference to which, as he said, would “turn the normal construction of criminal statutes upside-down, replacing the doctrine of lenity with a doctrine of severity.” Id., at 178. Here, the Nebraska Attorney General has adopted a narrow view of a criminal statute, one that comports with the rule of lenity (not to mention the statute’s plain meaning).
1 read the majority opinion to concede, if only implicitly, that the State has a legitimate interest in banning this dehumanizing procedure. The threshold question under Casey is whether the abortion regulation serves a legitimate state interest. 505 U. S. 833 (1992). Only if the statute serves a legitimate state interest is it necessary to consider whether the regulation imposes a substantial obstacle to women seeldng an abortion. Ibid. The fact that the majority considers whether Nebraska’s statute creates a substantial obstacle suggests that the Members of the majority other than Justice Stevens and Justice Ginsbukg have rejected respondent’s threshold argument that the statute serves no legitimate state purpose.
Justice Ginsburg seems to suggest that even if the Nebraska statute does not impose an undue burden on women seeking abortions, the statute is unconstitutional because it has the purpose of imposing an undue burden. Justice Ginsburg’s view is, apparently, that we can presume an unconstitutional purpose because the regulation is not designed to save any fetus from “destruction” or protect the health of pregnant women and so must, therefore, be designed to “chip away at... Roe.” Ante, at 952. This is a strange claim to make with respect to legislation that was enacted in 30 individual States and was enacted in Nebraska by a vote of 45 to 1, Nebraska Legislative Journal, 95th Leg., 1st Sess., 2609 (1997). Moreover, in support of her assertion that the Nebraska Legislature acted with an unconstitutional purpose, Justice Ginsburg is apparently unable to muster a single shred of evidence that the Nebraska legislation was enacted to prevent women from obtaining abortions (a purpose to which it would be entirely ineffective), let alone the kind of persuasive proof we would require before concluding that a legislature acted with an unconstitutional intent. In fact, as far as I can tell, Justice Ginsburg’s views regarding the motives of the Nebraska Legislature derive from the views of a dissenting Court of Appeals judge discussing the motives of legislators of other States. Justice Ginsburg’s presumption is, in addition, squarely inconsistent with Casey, which stated that States may enact legislation to “express profound respect for the life of the unborn,” 505 U. S., at 877, and with our opinion in Mazurek v. Armstrong, 520 U. S. 968 (1997) (per curiam), in which we stated:
“[E]ven assuming... that a legislative purpose to interfere with the constitutionally protected right to abortion without the effect of interfering with that right . . . could render the Montana law invalid — there is no basis for finding a vitiating legislative purpose here. We do not assume unconstitutional legislative intent even when statutes produce harmful results, see, e. g., Washington v. Davis, 426 U. S. 229, 246 (1976); much less do we assume it when the results are harmless.” Id., at 972 (emphases in original).
As I discuss below, the only question after Casey is whether a ban on partial birth abortion without a health exception imposes an “undue burden” on a woman seeking an abortion, meaning that it creates a “substantial obstacle” for the woman. I assume that the Court does not discuss the health risks with respect to undue burden, and instead suggests that health risks are relevant to the necessity of a health exception, because a marginal increase in safety risk for some women is clearly not an undue burden within the meaning of Casey. At bottom, the majority is using the health exception language to water down Casey’s undue-burden standard.
The majority’s conclusion that health exceptions are required whenever there is any support for use of a procedure is particularly troubling because the majority does not indicate whether an exception for physical health only is required, or whether the exception would have to account for “all factors — physical, emotional, psychological, familial, and the woman’s age — relevant to the well being of the patient.” Doe v. Bolton, 410 U. S. 179, 192 (1973). See also Voinovich v. Women’s Medical Professional Corp., 523 U. S. 1036, 1037 (1998) (Thomas, J., joined by Rehnquist, C. J., and Scaua, J., dissenting from denial of certiorari).
Use of the procedure may increase the risk of complications, including cervical incompetence, because it requires greater dilation of the cervix than other forms of abortion. See Epner, Jonas, & Seckinger, Late-term Abortion, 280 JAMA 724, 726 (Aug. 26, 1998). Physicians have also suggested that the procedure may pose a greater risk of infection. See Planned Parenthood of Wis. v. Doyle, 44 F. Supp. 2d 975, 979 (WD Wis. 1999). See also Sprang & Neerhof, Rationale for Banning Abortions Late in Pregnancy, 280 JAMA 744 (Aug. 26, 1998) (“Intact D&X poses serious medical risks to the mother”).
Nebraska was entitled to rely on testimony and evidence presented to Congress and to other state legislatures. Cf Erie v. Pap’s A. M., 529 U. S. 277, 296-297 (2000); Renton v. Playtime Theatres, Inc., 475 U. S. 41, 51 (1986). At numerous points during the legislative debates, various members of the Nebraska Legislature made clear that that body was aware of, and relying on, evidence before Congress and other legislative bodies. See App. in Nos. 98-3245, 98-3300 (CA8), pp. 846, 852-853, 878-879, 890-891, 912-913.