Background. Food frequency questionnaires (FFQs) have been validated in pregnant women, but few studies have focused specifically on low-income women and minorities. The purpose of this study was to examine the validity of the Harvard Service FFQ (HSFFQ) among low-income American Indian and Caucasian pregnant women. Methods. The 100-item HSFFQ was administered three times to a sample of pregnant women, and two sets of 24-hour recalls (six total) were collected at approximately 12 and 28 weeks of gestation. The sample included a total of 283 pregnant women who completed Phase 1 of the study and 246 women who completed Phase 2 of the study. Deattenuated Pearson correlation coefficients were used to compare intakes of 24 nutrients estimated from the second and third FFQ to average intakes estimated from the week-12 and week-28 sets of diet recalls. Results. Deattenuated correlations ranged from 0.09 (polyunsaturated fat) to 0.67 (calcium) for Phase 1 and from 0.27 (sucrose) to 0.63 (total fat) for Phase 2. Average deattenuated correlations for the two phases were 0.48 and 0.47, similar to those reported among other groups of pregnant women. Conclusion. The HSFFQ is a simple self-administered questionnaire that is useful in classifying low-income American Indian and Caucasian women according to relative dietary intake during pregnancy. Its use as a research tool in this population may provide important information about associations of nutrient intakes with pregnancy outcomes and may help to identify groups of women who would benefit most from nutritional interventions.