Studies of reproductive hazards in the workplace must address potential biases related to selection for employment. The National Natality Survey, a probability sample of live births to married women in 1980, was used to examine the relation between female employment during pregnancy and factors that might affect reproductive outcome by analyzing the 5,927 women with complete occupational data. Demographic and behavioral attributes as well as reproductive history were compared for the 3,712 women employed and the 2,215 women not employed during pregnancy. Employed mothers were of more optimal reproductive age, were more highly educated, had higher incomes, began prenatal care earlier, had greater weight gain during pregnancy, and were slightly less likely to be heavy smokers. Employed women had markedly fewer previous births and less favorable reproductive histories (more stillbirths, miscarriages, and induced abortions) than unemployed women, controlling for gravidity. Full- and part-time workers were similar with regard to demographic and behavioral characteristics, but part-time workers had higher parity. Differences were noted by employment sector: professional women had especially favorable demographic and behavioral traits, and women employed as operatives and service workers were less advantaged. These results indicate that substantial differences in pregnancy-related risk factors exist in relation to employment, with working women generally having more favorable demographic and behavioral characteristics and less favorable reproductive histories. This pattern could produce selection bias in studies of work and reproductive health, and it encourages the restriction of comparison groups to other employed women, with a need to consider heterogeneity among working women as well.