Ovulatory function was prospectively assessed over 6 mo in 23 vegetarians and 22 nonvegetarians with clinically normal menstrual cycles. Subjects were 20-40 y of age, of stable weight (body mass index, in kg/m2, of 18-25), on current diets for > or = 2 y, and not using oral contraceptives. Quantitative analysis of basal body temperature records classified cycles as normally ovulatory, short luteal phase (< 10 d), or anovulatory. Subjects completed the Three-Factor Eating Questionnaire (subjects completed the Three-Factor Eating Questionnaire (subscales for restraint, hunger, and disinhibition) and kept three 3-d food records. Vegetarians had lower BMIs (21.1 +/- 2.3 vs 22.7 +/- 1.9, P < 0.05), percentage body fat (24.0 +/- 5.5% vs 27.4 +/- 5.1%, P < 0.05), and restraint scores (6.4 +/- 4.4 vs 9.5 +/- 3.7, P < 0.05). Mean cycle lengths were similar, but vegetarians had longer luteal phase lengths (11.2 +/- 2.6 vs 9.1 +/- 3.8 d, P < 0.05). Cycle types also differed (chi 2 = 9.64, P < 0.01): vegetarians had fewer anovulatory cycles (4.6% vs 15.1% of cycles). Compared with those with restraint scores below the median, highly restrained women had fewer ovulatory cycles (3.6 +/- 2.3 vs 5.0 +/- 1.4, P < 0.05) and shorter mean luteal phase lengths (7.4 +/- 4.1 vs 10.7 +/- 3.1 d, P < 0.05). We conclude that ovulatory disturbances and restrained eating are less common among vegetarians, and that restraint influences ovulatory function.