Popper's proposal that scientific statements are distinguished by their falsifiability retains an appeal that its philosophical fate does not justify. The criterion is clean, it explains the discomfort many feel about theories that accommodate every outcome, and it can be stated in a sentence. Its difficulty is that no hypothesis confronts evidence alone: a prediction is derived from the hypothesis together with assumptions about instruments, background conditions and auxiliary theories, so a failed prediction indicts the conjunction and leaves the choice of what to abandon undetermined. Historical episodes cited as exemplary falsifications turn out, on examination, to involve exactly this kind of judgement. Falsifiability survives as a description of an attitude — a willingness to specify in advance what would count against one's view — rather than as a rule that sorts statements into two boxes.