Our study tests several hypotheses concerning the effects of employment, marriage, and motherhood on women's general physical health. These hypotheses predict how the health effect of each role varies, depending on specific role characteristics and the other roles a woman holds. Our analyses utilize longitudinal panel data for 3,331 women from the National Longitudinal Surveys of Young Women (follow-up intervals: 1978-83 and 1983-88). The Role Substitution Hypothesis proposes that employment and marriage provide similar resources (e.g., income and social support), and consequently, employment and marriage can substitute for each other in their beneficial effects on health. As predicted, we found that employment had beneficial effects on health for unmarried women, but little or no effect for married women. Similarly, marriage had beneficial effects on health only for women who were not employed. The Role Combination Strain Hypothesis proposes that employed mothers experience role strain, resulting in harmful effects on health. However, we found very little evidence that the combination of employment and motherhood resulted in harmful health effects. Contrary to the predictions of the Quantitative Demands Role Strain Hypothesis, it appears that neither longer hours of employment nor having more children resulted in harmful effects on health. As predicted by the Age-Related Parental Role Strain Hypothesis, younger age at first birth, particularly a teenage birth, appeared to result in more harmful health effects.