Sickness absence in pregnancy has been shown to be associated with strenuous working conditions and parity. So far, few studies have made adjustments for possible interaction and confounding. Such adjustments are needed to more precisely identify targets for preventive measures. We have, therefore, in a representative population of pregnant employees in Norway 1989, computed adjusted odds ratios for leaving work by sickness absence more than three (LSC > 3) and eight (LSC > 8) weeks before delivery according to working conditions identified as risk factors in earlier studies; adjusted for job control, domestic conditions and sickness absence the year prior to pregnancy. The cumulative percentage of LSC > 8 and LSC > 3 was 26.4 and 51.1. Ergonomically strenuous postures and heavy lifting increased the risk of both outcomes. In addition, shift work and hectic work pace increased the risk of LSC > 3. Influence on breaks reduced risk. Only para experienced reduced risk of LSC when working part-time. Sicklisting the year prior to pregnancy had no confounding effect, which suggest that pregnancy represents a new incompatibility with work. Preventive measures should address work postures and heavy lifting, as well as conditions influencing the woman's control with her time.