HIV serostatus and menstrual function were examined using prospectively collected menstrual data from 802 HIV-seropositive and 273 HIV-seronegative women, ages 20 to 44, enrolled in two cohort studies of HIV infection in North American women. The associations between HIV serostatus and the probabilities of having a cycle lasting >40 days (n = 541 cycles), >90 days (n = 67 cycles), <18 days (n = 316 cycles) and mean length and variability of 18 to 40 day cycles (n = 3634) were assessed. After adjustment for demographic characteristics, body mass index, and substance use, seropositivity increased the odds of having a very short cycle (<18 days, odds ratio [OR], 1.45; 95% confidence interval [CI], 1.00-2.11) and a very long cycle (>90 days, OR, 1.32; 95% CI, 0.68-2.58) slightly, although the latter CIs include one. Seropositivity did not increase the odds of having a moderately long cycle (>40 days, OR, 1.14) or affect mean cycle length or variability (β, 0.30 ± 0.20; between-woman standard deviation [SD], 2.2 days [HIV-seronegative] and 1.9 days [HIV-seropositive]; within-woman SD, 3.5 days for both). Although seropositivity may slightly increase the probability of very short cycles, HIV serostatus has little overall effect on amenorrhea, menstrual cycle length, or variability. Among HIV-seropositive women, higher viral loads and lower CD4+ counts were associated with increased cycle variability and polymenorrhea.