Studies of occupational exposure and spontaneous abortion may use pregnancies during which the mother was unemployed as part or all of the unexposed comparison group. Any type of maternal employment, however, may be a risk factor for spontaneous abortion, and potential confounder in occupational reproductive studies. This study evaluates the effect of employment in a cohort of pregnancies of 1535 women. Employed pregnancies had a significantly higher rate of spontaneous abortion (14.5%) than unemployed pregnancies (11.7%) (RR = 1.23, 95% CI = 1.02, 1.49). Gravidity acted as an effect modifier, as the employment effect was seen only in multigravidous pregnancies (RR = 1.38, 95% CI = 1.11, 1.72) and not primigravidous pregnancies (RR = 0.96). The effect persisted when an independent sample of one randomly selected pregnancy per woman was used for the analysis (RR = 1.27, 95% CI = 0.90, 1.79). The data were examined for confounding by other factors which could explain the excess in spontaneous abortion among employed pregnancies. The employment effect persisted with adjustment for other risk factors including maternal age, education, income, maternal diabetes, race, alcohol usage and smoking, and prior pregnancy ending in induced abortion. Stratifying by prior pregnancy loss eliminated the employment effect among those with prior loss (RR = 1.03) but enhanced the effect among those multigravidous without the risk factor (RR = 1.50, 95% CI = 1.15, 1.97). Selection bias, also, was explored as a possible explanation of this employment effect, but could not be substantiated. Assessment of a true exposure effect requires consideration of a potential employment effect either in the design or analysis.