The claim that the best language-teaching materials are based on a contrast of the two competing linguistic systems has long been a popular one in language teaching. It exists in strong and weak versions, the strong one arising from evidence from the availability of some kind of metatheory of contrastive analysis and the weak from evidence from language interference. The strong version of the hypothesis is untenable and even the weak version creates difficulties for the linguist. Recent advances in linguistic theory have led some people to claim that the hypothesis is no longer useful in either the strong or the weak version. Such a claim is perhaps unwarranted, but a period of quiescence is probable for contrastive analysis itself.