I use Danish survey and administrative data to examine the impact of maternal employment during pregnancy on birth outcomes. As healthier mothers are more likely to work and health shocks to mothers may impact employment and birth outcomes, I combine two strategies: First, I control extensively for time-varying factors that may correlate with employment and birth outcomes, such as pre-pregnancy family income and maternal occupation, pregnancy-related health shocks, maternal sick listing, and health behaviors (smoking and alcohol consumption). Second, to account for remaining time-invariant heterogeneity between mothers, I compare outcomes of mothers' consecutive children. Mothers who work during the first pregnancy trimester have a lower risk of preterm birth. I find no effect on the probability of having a baby of small size for gestational age. To rule out that health selection of mothers between pregnancies drives the results, I focus on mothers whose change in employment status is likely not to be driven by underlying health (mothers who are students in one of their pregnancies and mothers with closely spaced births). Given generous welfare benefits and strict workplace regulations in Denmark, my findings support a residual explanation, namely, that exclusion from employment may stress mothers in countries with high-female employment rates.