OBJECTIVE: This study compares gestational age data obtained by clinical estimate with data calculated from the date of the last menstrual period (LMP) as recorded on birth certificates. METHODS: The authors analyzed 476,034 computerized birth records from three overlap years, that is, those that contained both menstrual and clinical estimates of gestational age, concentrating on cases within the biologically plausible range of 20-44 weeks. RESULTS: The overall exact concordance between the two measurements was 46%. For +1 week it was 78%, and for +2 weeks it was 87%. Incidence of prematurity was 16% with menstrual gestational age, while it was 12% with clinical estimate. About 47% of the LMP-based preterm births were classified as term by clinical estimate. Eighty-three percent of clinical estimate-based preterms were also preterms by LMP-based gestation. Birthweight frequency distribution curves for LMP-based gestational age are bimodal, indicating probable miscoding of term births. An apparent over-representation of births coded as exactly 40 weeks by clinical estimate suggests rounding off near term for this method. CONCLUSION: Agreement between menstrual and clinical estimates of gestational age occurs most often close to term, with significant disagreement in preterm and postterm births. Use of different methods of determining gestation in different years or geographic populations will result in artifactual differences in important indicators such as prematurity rate.