OBJECTIVES: To examine differences in the pattern of weight changes during and after pregnancy among four pregravid body mass index (BMI) groups. STUDY DESIGN: Prospective cohort study of women who had two consecutive births at the University of California, San Francisco (UCSF) between 1980 and 1990. MEASUREMENTS: Maternal body weights were available before conception and delivery, and at 6 weeks postpartum for the first (index) pregnancy, and before conception for the second study pregnancy. Height and two pregravid weights were self-reported. Weights at delivery and 6 weeks postpartum were measured. Net delivery weight was defined as delivery weight minus infant birth weight. Three non-overlapping sequential weight changes were constructed: (1) net gestational gain (net delivery weight minus pregravid weight at the index pregnancy); (2) early net postpartum weight change (6-week postpartum weight minus net delivery weight); and (3) late postpartum weight change (pregravid weight at the second pregnancy minus 6-week postpartum weight). SUBJECTS: A total of 985 healthy women (age 18-41 y) from four race/ethnicity groups (Asian, Hispanic, black and white) who had a singleton, full-term, live birth for the index pregnancy followed by a second consecutive birth. RESULTS: Four race/ethnicity groups were combined (no interaction) to contrast average weight changes among pregravid BMI groups. Means adjusted for eight covariates (parity, race/ethnicity, education, mode of delivery, smoking, hypertension of pregnancy, age, height) and time intervals were not altered appreciably. Early net postpartum weight losses were similar for all pregravid BMI groups. Late (median of 2 y) postpartum weight losses were 4 kg higher in the low and average BMI groups compared with the highest BMI group. About half of the net gestational gain was lost by 6 weeks postpartum, and the percentage that was lost decreased over time. CONCLUSIONS: This study suggests that early postpartum weight loss does not vary by maternal pregravid BMI group, but late postpartum weight change does. Serial weight measurements are needed in epidemiologic studies to differentiate retention of gestational gain from weight gain during the late postpartum period.