Using data on the number of men and women who received doctorates in all academic fields from 1971 to 2002, the authors examine changes in the sex composition of fields. During this period, the proportion of women who received doctorates increased dramatically from 14 percent to 46 percent. Regression models with fixed effects indicate no evidence that fields with declining relative salaries deter the entry of men, as would be predicted by the queuing theory of Reskin and Roos. Consistent with the devaluation perspective and Schelling's tipping model, above a certain percentage of women, men are deterred from entering fields by the fields' further feminization. However, the rank order of fields in the percentage of women changed only slightly over time, implying that, to a large extent, men and women continued to choose fields as before, even when many more women received doctorates. The findings on the effects of feminization on salaries are mixed.