with whom The Chief Justice and Justice White join, dissenting.
Many people come to our country because they fear persecution in their homeland. Congress has provided two forms of relief for such people: asylum, see Immigration and Nationality Act of 1952, § 208(a), as added by 94 Stat. 105, 8 U. S. C. § 1158(a); and withholding of deportation, see 66 Stat. 212, § 248(h), as amended, 94 Stat. 107, 8 U. S. C. § 1253(h). The Board of Immigration Appeals (BIA) has concluded that there is no practical distinction between the objective proofs an alien must submit to be eligible for these two forms of relief. The Court rejects this conclusion. Because I believe the BIA’s interpretation of the statute is reasonable, I dissent.
I
The Court’s opinion seems to assume that the BIA has adopted a rigorous mathematical approach to asylum cases, requiring aliens to demonstrate an objectively quantifiable risk of persecution in their homeland that is more than 50%. The Court then argues that such a position is inconsistent with the language and history of the Act. But this has never been the BIA’s position. Thus, it is useful to examine the BIA’s approach in some detail before evaluating the Court’s *456rejection of the BIA’s approach. After all, the BIA is the tribunal with the primary responsibility for applying the Act and the greatest experience in doing so.
The BIA’s interpretation of the statutory term “well-founded fear” appears in Matter of Acosta, Interim Decision No. 2986 (Mar. 1, 1985).1 Under the BIA’s analysis, an immigration judge evaluating an asylum application should begin by determining the underlying historical facts. The burden of persuasion rests on the applicant, who must establish the truth of these facts by a preponderance of the evidence. See id., at 7 (citing, inter alia, 1A C. Gordon & H. Rosenfield, Immigration Law and Procedure §5.10b, p. 5-121 (rev. ed. 1986)).
Once the immigration judge has decided what historical facts the applicant has demonstrated, he then decides whether those facts meet the definition of “refugee” set forth in § 101(a)(42)(A) of the Act, 8 U. S. C. § 1101(a)(42)(A). The major point of contention in this case concerns that section’s requirement that the fear be “well-founded.”2 In *457Acosta, the BIA adhered to the interpretation of that language it had developed in Matter of Dunar, 14 I. & N. Dec. 310 (1973):
“ ‘[T]he requirement that the fear be “well-founded” rules out an apprehension which is purely subjective.... Some sort of showing must be made and this can ordinarily be done only by objective evidence. The claimant’s own testimony as to the facts will sometimes be all that is available; but the crucial question is whether the testimony, if accepted as true, makes out a realistic likelihood that he will be persecuted.”’ Acosta, supra, at 18-19 (quoting Dunar, supra, at 319) (emphasis added by Acosta Board).
The Acosta Board went on to caution:
“By use of such words [as ‘realistic likelihood’] we do not mean that ‘a well-founded fear of persecution’ requires an alien to establish to a particular degree of certainty, such as a ‘probability’ as opposed to a ‘possibility,’ that he will become a victim of persecution. Rather as a practical matter, what we mean can best be described as follows: the evidence must demonstrate that (1) the alien possesses a belief or characteristic a persecutor seeks to overcome in others by means of punishment of some sort; (2) the persecutor is already aware, or could easily become aware, that the alien possesses this belief or characteristic; (3) the persecutor has the capability of punishing the alien; and (4) the persecutor has the inclination to punish the alien.” Acosta, supra, at 22.
Finally, the Acosta opinion compared this “realistic likelihood” standard to the “clear probability” standard applied to *458applications for withholding of deportation. The BIA’s comments are insightful:
“One might conclude that ‘a well-founded fear of persecution,’ which requires a showing that persecution is likely to occur, refers to a standard that is different from ‘a clear probability of persecution,’ which requires a showing that persecution is ‘more likely than not’ to occur. As a practical matter, however, the facts in asylum and withholding cases do not produce clear-cut instances in which such fine distinctions can be meaningfully made. Our inquiry in these cases, after all, is not quantitative, i. e., we do not examine a variety of statistics to discern to some theoretical degree the likelihood of persecution. Rather our inquiry is qualitative: we examine the alien’s experiences and other external events to determine if they are of a kind that enable us to conclude the alien is likely to become the victim of persecution. In this context, we find no meaningful distinction between a standard requiring a showing that persecution is likely to occur and a standard requiring a showing that persecution is more likely than not to occur. . . . Accordingly, we conclude that the standards for asylum and withholding of deportation are not meaningfully different and, in practical application, converge.” Id., at 25.
In sum, contrary to the Court’s apparent conclusion, the BIA does not contend that both the “well-founded fear” standard and the “clear probability” standard require proof of a 51% chance that the alien will suffer persecution if he is returned to his homeland. The BIA plainly eschews analysis resting on mathematical probabilities. Rather, the BIA has adopted a four-part test requiring proof of facts that demonstrate a realistic likelihood of persecution actually occurring. The heart of the Acosta decision is the BIA’s empirical conclusion, based on its experience in adjudicating asylum applications, that if the facts establish such a basis for an alien’s *459fear, it rarely will make a difference whether the judge asks if persecution is “likely” to occur or “more likely than not” to occur. If the alien can establish such a basis, he normally will be eligible for relief under either standard.
II
In Part II of its opinion, the Court examines the language of the Act. Section 243(h) provides that the Attorney General shall grant withholding of deportation to any country where “such alien’s life or freedom would be threatened.” 8 U. S. C. § 1253(h). . Section 208(a) provides that the Attorney General has discretion to grant asylum “if the Attorney General determines that such alien is a refugee.” § 1158(a). The crucial language of § 101(a)(42)(A) of the Act, as added by 94 Stat. 102, defines a refugee as a person who has “a well-founded fear of persecution.” § 1101(a)(42)(A). In the Court’s view, this language all but disposes of the case. Ante, at 427-432.
With respect to the issue presented by this case, I find the language far more ambiguous than the Court does. Respondent contends that the BIA has fallen into error by equating the objective showings required under §§ 208(a) and 243(h). The Court notes that the language of § 208(a) differs from the language of § 243(h) in that it contemplates a partially subjective inquiry. From this premise, the Court moves with little explanation to the conclusion that the objective inquiries under the two sections necessarily are different.
In reaching this conclusion, the Court gives short shrift to the words “well-founded,” that clearly require some objective basis for the alien’s fear. The critical question presented by this case is whether the objective basis required for a fear of persecution to be “well-founded” differs in practice from the objective basis required for there to be a “clear probability” of persecution. Because both standards necessarily contemplate some objective basis, I cannot agree with the Court’s *460implicit conclusion that the statute resolves this question on its face. In my view, the character of evidence sufficient to meet these two standards is a question best answered by an entity familiar with the types of evidence and issues that arise in such cases. Congress limited eligibility for asylum to those persons whom “the Attorney General determines” to be refugees. See § 208(a), 8 U. S. C. § 1158(a). The Attorney General has delegated the responsibility for making these determinations to the BIA. That Board has examined more of these cases than any court ever has or ever can. It has made a considered judgment that the difference between the “well-founded” and the “clear probability” standards is of no practical import: that is, the evidence presented in asylum and withholding of deportation cases rarely, if ever, will meet one of these standards without meeting both. This is just the type of expert judgment — formed by the entity to whom Congress has committed the question — to which we should defer.
The Court ignores the practical realities recognized by the expert agency and instead concentrates on semantic niceties. It posits a hypothetical situation in which a government sought to execute every 10th adult male. In its view, fear of such executions would be “well-founded” even if persecution of a particular individual would not be “more likely than not” to occur. See ante, at 431. But this hypothetical is irrelevant; it addresses a mathematically demanding interpretation of “well-founded” that has no relation to the BIA’s actual treatment of asylum applications. Nor does it address the validity of the BIA’s judgment that evidence presenting this distinction will be encountered infrequently, if ever.
Common sense and human experience support the BIA’s conclusion. Governments rarely persecute people by the numbers. It is highly unlikely that the evidence presented at an asylum or withholding of deportation hearing will demonstrate the mathematically specific risk of persecution posited by the Court’s hypothetical. Taking account of the *461types of evidence normally available in asylum cases, the BIA has chosen to make a qualitative evaluation of “realistic likelihoods.” As I read the Acosta opinion, an individual who fled his country to avoid mass executions might be eligible for both withholding of deportation and asylum, whether or not he presented evidence of the numerical reach of the persecution. See Acosta, Interim Decision No. 2986, at 18-25.3 Nowhere does the Court consider whether the BIA’s four-element interpretation of “well-founded” is unreasonable. Nor does the Court consider the BIA’s view of the types of evidentiary presentations aliens generally make in asylum cases.
In sum, the words Congress has chosen — “well-founded” fear — are ambiguous. They contemplate some objective basis without specifying a particular evidentiary threshold. There is no reason to suppose this formulation is inconsistent with the analysis set forth in Acosta. The BIA has concluded that a fear is not “well-founded” unless the fear has an objective basis indicating that there is a “realistic likelihood” that persecution would occur. Based on the text of the Act alone, I cannot conclude that this conclusion is unreasonable.
I — I I — I ) — I
The Court bolsters its interpretation of the language of the Act by reference to three parts of the legislative history. A closer examination of these materials demonstrates that each of them is ambiguous. Nothing the Court relies on provides a positive basis for arguing that there is a material difference between the two standards.
*462A
First, the Court cites legislative history indicating that Congress wished to preserve some existing standard when it placed the words “well-founded fear” in the Act. The Court concludes that the standard Congress intended to preserve was the BIA’s practice under the old § 203(a)(7), 79 Stat. 913 (1965). That section authorized the Attorney General to grant conditional entry to aliens fleeing from Communist countries or the Middle East, so long as they established a “fear of persecution.” The Court argues that Congress chose the words “well-founded fear” to “preserve” as an asylum standard the prior interpretation of the word “fear” in the standard for conditional entry.
In contrast, the United States argues that Congress chose the words “well-founded fear” to preserve the Attorney General’s regulations governing applications for asylum by aliens in the United States.4 These regulations were substantially in accord with the BIA’s view, namely that there is no significant difference between the “well-founded fear” and “clear probability” standards. Compare 8 CFR §§ 108.3(a) and 236.3 (a)(2) (1980) (asylum) with 8 CFR §242.17(c) (1980) (withholding of deportation). Common sense suggests that the United States has the better of this argument. It is more natural to speak of “preserving” an interpretation that had governed the same form of relief than one that had applied to a different form of relief.
Moreover, the legislative history makes it clear that Congress was referring to the regulations rather than to § 203(a)(7). The Senate Report states that the bill “im-*463prov[es] and clarifies] the procedures for determining asylum claims filed by aliens who are physically present in the United States. The substantive standard is not changed.” S. Rep. No. 96-256, p. 9 (1979). As the Court recognizes, ante, at 435, n. 17, this statement unquestionably refers to the informal procedures for aliens in the United States, not the statutory procedures under § 203(a)(7).5 Similarly, the House Report states that “the new definition does not create a new and expanded means of entry, but instead regularizes and formalizes the policies and practices that have been followed in recent years.” H. R. Rep. No. 96-608, p. 10(1979) (emphasis added). Congress hardly would have felt a need to “formalize” the statutory procedures under § 203(a)(7). Indeed, the House Report cites the Attorney General’s regulations as the extant procedures to which it was referring. H. R. Rep., at 17.
In my view, the legislative history indicates that Congress’ choice of the words “well-founded” fear as the standard of eligibility for asylum was intended to carry forward the practice of the Attorney General in adjudicating asylum applications. The Attorney General had concluded that the standard for asylum was substantially identical to the standard for withholding of deportation. His decision to interpret the language of § 208 in the same way is entirely reasonable.
B
Second, the Court relies on materials interpreting the United Nations Protocol. Ante, at 437-440. For several reasons, I find these materials to be only marginally rele*464vant. Both the President and the Senate thought that the Protocol was perfectly consistent with our country’s immigration laws. See INS v. Stevic, 467 U. S. 407, 417 (1984) (citing legislative history). We should be reluctant to assume that our country has been violating the Protocol during the 20 years since its adoption. Moreover, as the Court recognizes, statements by the United Nations High Commissioner for Refugees have no binding force, because “‘the determination of refugee status under the . . . Protocol... is incumbent upon the Contracting State.’” Ante, at 439, n. 22 (quoting Office of the United Nations High Commissioner for Refugees, Handbook on Procedures and Criteria for Determining Refugee Status l(ii) (Geneva, 1979)).
In any event, the materials discussed by the Court shed little or no light on the question presented by this case. None of them states that the burden of proof for nonrefoulement under Article 33.1 of the United Nations Protocol of 1967 — a remedy essentially identical to withholding of deportation under § 243(h) of the Act — is higher than the burden of proof for asylum under Article 34. The only thing the materials tend to establish is that a mathematical approach to the likelihood of persecution in asylum cases is arguably inconsistent ■with the sense of the drafters of the Protocol. The BIA has declined to adopt such an approach. See supra, at 457-459. It is simply irrelevant that this approach might be inconsistent with the views of commentators on the Protocol.
C
Finally, the Court places great weight on the changes in the Act made by the Conference Committee. The Court notes that the Senate bill, S. 643, authorized the Attorney General to grant asylum if the applicant “is a refugee within the meaning of section 101(a)(42)(A) and his deportation or return would be prohibited under section 243(h) of this Act.” S. Rep. No. 96-256, at 26. The Court conjectures that this language “indicates that the Senate recognized that *465there is a difference between the ‘well-founded fear’ standard and the clear-probability standard. The enactment of the House bill rather than the Senate bill in turn demonstrates that Congress eventually refused to restrict eligibility for asylum only to aliens meeting the stricter standard.” Ante, at 442 (footnote omitted).
Neither the premise of the Court nor its conclusion is justified. The language of the Senate bill does not demonstrate that the Senate recognized a difference between the two standards. The Senate just as easily could have included the language to ensure that the Attorney General held to his position that there was no difference between the standards. Moreover, there is no reason to believe that the changes made by the Conference Committee reflected a considered rejection of this portion of the Senate’s definition of refugee. Rather, the Conference Committee Report demonstrates that the Conference thought both bills adopted the same general definition of refugee — the U. N. definition. See H. R. Conf. Rep. No. 96-781, p. 19 (1980). The differences the Conference saw between the bills related to treatment of refugees still in their homeland, and to refugees who have been “firmly resettled” in another country. See ibid.
In short, I see no reason to believe that the minor differences in wording between the Senate bill and the Act as passed reflect a rejection of the position that there is no significant difference between the two standards.6 Thus, I place no weight on the Conference Committee’s choice of the language of the House bill.
IV
Even if I agreed with the Court’s conclusion that there is a significant difference between the standards for asylum and *466withholding of deportation, I would reverse the decision of the Court of Appeals and uphold the decision of the BIA in this case.7 A careful reading of the decisions of the BIA and the Immigration Judge demonstrates that the BIA applied the lower asylum standard to this case.
Respondent’s claim for asylum rested solely on testimony that her brother had experienced difficulties with the authorities in Nicaragua. The Immigration Judge rejected respondent’s claim because he found “no evidence of any substance in the record other than her brother’s claim to asylum.” App. to Pet. for Cert. 27a. He further found:
“None of the evidence indicates that the respondent would be persecuted for political beliefs, whatever they may be, or because she belongs to a particular social group. She has not proven that she or any other members of her family, other than her brother, has [sic] been detained, interrogated, arrested and imprisoned, tortured and convicted and sentenced by the regime presently in power in Nicaragua.” Ibid.
The absence of such evidence was particularly probative, because many of the other members of respondent’s family— her parents, two sisters, her brother’s wife, and her broth*467er’s two children — were still in Nicaragua and thus presumably subject to the persecution respondent feared.
On appeal, the BIA affirmed. It decided this case after the passage of the Act, but before its opinion in Acosta. At that time, the BIA was confronted with a number of conflicting decisions by Courts of Appeals as to the correct standard for evaluating asylum applications. The BIA noted three different formulations of the “well-founded fear” standard: the “clear probability” test, see Rejaie v. INS, 691 F. 2d 139 (CA3 1982); the “good reason” test, see Stevic v. Sava, 678 F. 2d 401 (CA2 1982), rev’d on other grounds, INS v. Stevic, 467 U. S. 407 (1984); and the “realistic likelihood” test the BIA had adopted in Matter of Dunar, 141. & N. Dec. 310 (1973). App. to Pet. for Cert. 21a. See supra, at 456-469 (discussing Acosta). Reviewing the evidence respondent had submitted to the Immigration Judge, the BIA concluded that respondent could not obtain relief under any of the standards. The BIA focused especially on the fact that respondent
“has openly admitted that she herself has taken no actions against the Nicaraguan government. She admits that she has never been politically active. She testified that she never assisted her brother in any of his political activities. Moreover, she admits that she has never been singled out for persecution by the present government.” App. to Pet. for Cert. 22a.8
Respondent filed a petition for review with the Court of Appeals for the Ninth Circuit. Without examining either the factual or legal basis for the BIA’s decision, the court granted the petition, reversed the BIA’s decision, and remanded the application to the BIA for further consideration. *468767 F. 2d 1448 (1985). The sole basis articulated for this action was a conclusion that the BIA had applied the wrong legal standard. The Court of Appeals repeated its position that the standards for asylum and withholding of deportation are different. According to that court, an asylum applicant must “present ‘specific facts’ through objective evidence to prove either past persecution or ‘good reason’ to fear future persecution.” Id., at 1453 (quoting Carvajal-Munoz v. INS, 743 F. 2d 562, 574 (CA7 1984)) (emphasis added). It then noted that the BIA had reached a different conclusion in Acosta and stated:
“[T]he Board appears to feel that it is exempt from the holding of Marbury v. Madison . . . and not constrained by circuit court opinions. . . . [T]he Board applied its own construction of the applicant’s burden of proof in an asylum case to the claims of both Cardoza-Fonseca and [her copetitioner]. It held that they were required to demonstrate a clear probability of persecution in order to be declared eligible for asylum.” 767 F. 2d, at 1454 (citation omitted).
This statement is simply inconsistent with the BIA’s opinion. As I have explained, the BIA acknowledged the conflicting decisions of the various Courts of Appeals and explicitly tested the application under three different standards. The least burdensome of these — the “good reason” standard — is identical to the court’s statement quoted supra this page. The Court of Appeals completely ignored the words in which the BIA framed its decision. It failed to examine the factual findings on which the decision rested. At least in this case, it appears that the Court of Appeals, and not the BIA, has misunderstood the proper relation between courts and agencies. That court properly could have considered whether substantial evidence supported the BIA’s conclusion that respondent failed to demonstrate a “good reason” to fear persecution, but it should not have assumed that *469the BIA tested respondent’s application by a higher standard than the BIA’s own opinion reflects.
V
In my view, the Court misconstrues the Act and misreads its legislative history. Moreover, neither this Court nor the Court of Appeals has identified an error in the decision of the BIA in this case. Neither court has examined the factual findings on which the decision rested, or the legal standard the BIA applied to these facts. I would reverse the decision of the Court of Appeals.
The Court suggests that the BIA’s interpretation of the “well-founded fear” standard has been “erratic.” Ante, at 446-447, n. 30. An examination of the relevant BIA decisions leads to a contrary conclusion. The BIA first addressed the standard in Matter of Dunar, 14 I. & N. Dec. 310 (1973). In that ease, the BIA considered the meaning of the term “well-founded fear” in the United Nations Protocol Relating to the Status of Refugees, Jan. 31, 1967, 19 U.S.T. 6223, 6225, T.I.A.S. No. 6577 (1968). When Congress inserted this language in the asylum provisions of the Act in 1980, the BIA interpreted the language to mean exactly the same thing as the language in the Protocol. Matter of Acosta, Interim Decision No. 2986 (Mar. 1,1985). Thus, the BIA’s position has never changed. The Court bases its characterization of the BIA’s record on decisions applying the more lenient “fear” standard. If anything about these statutes is clear, it is that a “well-founded fear” is something more than a “fear.” It is unfair to characterize the BIA’s decisions as “erratic” when the agency was in fact interpreting two different standards.
The BIA has interpreted the statutory definition to require proof of four elements: (i) the alien must have a “fear” of “persecution”; (ii) the fear must be “well-founded”; (iii) the persecution must be “on account of race, *457religion, nationality, membership in a particular social group, or political opinion”; and (iv) the alien must be unable or unwilling to return to his homeland because of persecution or his well-founded fear of persecution. See id., at 11.
Of course, the applicant would have to meet all four elements of the well-founded fear standards. See supra, at 457 (quoting Acosta, Interim Decision No. 2986, at 22). Although these requirements restrict grants of relief in some cases, none of them rests on the mathematical, considerations that the Court suggests govern current BIA practice. Moreover, the Court’s exegesis of the “plain meaning” of the phrase “well-founded” in no way suggests that the BIA’s four-part test is a misinterpretation of the statute.
Those regulations constituted this country’s informal attempt to comply with the exhortation of the Convention Relating to the Status of Refugees to “facilitate the assimilation and naturalization,” Art. 34, Jan. 31, 1967, [1968] 19 U.S.T. 6259, 6276, T.I.A.S. No. 6577, of persons who have a “well-founded fear of being persecuted,” Art. 1(A)(2), id., at 6261. All parties agree that the Convention’s language was the ultimate source of the language Congress placed in the Act.
The Court concludes that the Senate Report has no probative force because the Conference Committee adopted the House language rather than the Senate language. But the changes in language made by the Conference Committee do not help the Court’s position. As I explain infra this page, the House Report indicates that the House bill also was intended to adopt the standards set forth in the regulations. Moreover, there is no suggestion in the Conference Report that this change in language affected the substantive standard. See infra, at 464-465.
This interpretation is supported by evidence that the House bill, like the Senate bill, was intended to preserve the Attorney General’s regulations treating the two standards as substantially identical. See supra, at 463.
The Court contends that this question is not before us. Ante, at 448, n. 31. I find this suggestion quite strange. The Immigration and Naturalization Service asked the Court to determine “[wjhether an alien’s burden of proving eligibility for asylum... is equivalent to his burden of proving eligibility for withholding of deportation.” Pet. for Cert. (I). The question whether the two standards are equivalent “fairly includes,” under this Court’s Rule 21.1(a) the problem of defining the appropriate standard for asylum. And that question can only be answered on the facts of this case. The Court does not sit to answer hypothetical questions of statutory construction. Normally we resolve such questions only by examining the facts of the case before us. In this case, the Court affirms the Court of Appeals’ decision that the BIA required an intolerably high burden of proof in this case. Yet, like the Court of Appeals, the Court examines neither the facts of the ease before us nor the legal standard the BIA applied. In my view, Rule 21 does not contemplate this result.
In terms of the four-element Acosta test for well-founded fear, respondent’s claim would have failed both the first and the second elements. Respondent failed to show either that she “possesses a belief or characteristic the persecutor seeks to overcome” or that “the persecutor is already aware, or could easily become aware, that [she] possesses this belief or characteristic.” Acosta, Interim Decision No. 2986, at 22.