Bioethics increasingly relies on empirical research in order to resolve ethical issues. Bioethics has also, of late, begun reimagining the researcher-subject relationship to include new positive duties to subjects, such as providing ancillary care and offering to return research results. These trends have converged in the form of a growing stream of empirical studies of subjects’ preferences about being offered research results. The legal and ethical question at the intersection of these trends is how we should factor these preferences into research design.This short article in the leading bioethics journal takes a middle position on this question in the context of a proposed longitudinal study of genes, environment and health, in which the National Human Genome Research Institute hopes to enroll a cohort of 500,000, primarily through door-to-door recruitment. A survey found that most respondents would want to be offered the return of their individual genetic results. I reject both (1) the claim that evidence that subjects want to be offered results entails a duty by researchers to do so, but also (2) the paternalistic position that most individual results should not be returned.I reach the first conclusion by arguing that the subject-researcher relationship in this case should be viewed as a sort of arms-length donative contract: an agreement under which the donor-subject agrees to make a gift to the donee-researcher, who arguably holds the gift in trust for society (especially given that the research is government-funded) and does not promise anything directly in return. A contract’s terms need not, and rarely do, reflect the ideal preferences of either party, and researchers may make offers to potential research volunteers that reflect their own preferences, resource limitations, and conscientious beliefs about the wisdom of returning certain results. The acceptance of what often must be, for pragmatic reasons, a take-it-or-leave-it offer to volunteer is not invalidated, nor the ensuing research rendered unethical, because the volunteer’s gift is not “repaid” in the form of research results or, indeed, in any form (though subjects must be told results will not be disclosed). Moreover, returning individual results in an ethical manner can be costly, and it would be odd (and perhaps not in keeping with subjects’ overall preferences, which most surveys are not designed to elicit or construct) if researchers had a duty to subjects that undercut their ability to devote the maximum available resources to the research itself — the very end that should motivate the subjects to volunteer in the first place. Finally, even if we assume that subjects own or are otherwise entitled to their genetic information, surely such property can be alienated, or such entitlements waived, through the donative contract. A contract’s terms, however, must not sink to the level of unconscionability.Empirical data can, however, inform researchers who admirably want, and are able, to honor subjects’ preferences. Subjects have interests in receiving a much wider range of results than is often acknowledged, and most of the psychosocial risks they are said to face from disclosure remain speculative. Returning results may also help researchers recruit and retain subjects.