The claim that a treatment is futile is often used to justify a shift in the physician's ethical obligations to patients. In clinical situations in which non-futile treatments are available, the physician has an obligation to discuss therapeutic alternatives with the patient. By contrast, a physician is under no obligation to offer, or even to discuss, futile therapies. This shift is supported by moral reasoning in ancient and modern medical ethics, by public policy, and by case law. Given this shift in ethical obligations, one might expect that physicians would have unambiguous criteria for determining when a therapy is futile. This is not the case. Rather than being a discrete and definable entity, futile therapy is merely the end of the spectrum of therapies with very low efficacy. Ambiguity in determining futility, arising from linguistic errors, from statistical misinterpretations, and from disagreements about the goals of therapy, undermines the force of futility claims. Decisions to withhold therapy that is deemed futile, like all treatment choices, must follow both clinical judgments about the chance of success of a therapy and an explicit consideration of the patient's goals for therapy. Futility claims rarely should be used to justify a radical shift in ethical obligations.