The incidence rates of dysplasia and carcinoma in situ (CIS) of the cervix and vagina were determined in 3,980 young women exposed prenatally to diethylstilbestrol. Strict criteria were developed to minimize selection bias among the subset of 744 pairs of matched exposed and unexposed (control) cohort participants, all of whom were identified through review of prenatal obstetrical records. A high degree of compliance was achieved throughout the seven-year study period since in each group about 90% of the women remained as active participants, kept 77% of the annual anniversary examinations, and had separate Papanicolaou smears of the cervix and vagina performed in 99% of the anniversary examinations. The incidence rate for dysplasia and CIS was significantly higher in the women exposed to diethylstilbestrol than in those not exposed in the matched cohort (15.7 v 7.9 cases per 1,000 person-years of follow-up). The rates were higher in the exposed women if squamous metaplasia extended to the outer half of the cervix or onto the vagina. In other respects, the matched cohorts were strikingly similar.