To determine the association between maternal activity and pregnancy outcome, the authors investigated the separate influences of time, physical effort, and posture at work, both at a job and in the home, on birth weight. Study subjects were 1,507 of 1,889 women appearing consecutively for antenatal care at a district general hospital in inner London, England, in 1982-1984. Data were collected prospectively by interview and examination at several stages of pregnancy. Multivariable linear regression was used to distinguish associations with physical activity from confounding by other factors. The mean birth weight of infants born to women in full-time employment was 49 g less than that for births to women not in paid work (95 percent confidence interval (Cl) 1-97 g). However, the difference was due to confounding, and after adjustment, full-time employment was associated with a 12-g increase in birth weight (95 percent Cl -39 to 63 g). There was little difference in birth weight related to gestational stage at leaving work. Duration, physical effort required, and energy expenditure in paid work and in work at home had no discernible association with birth weight. A small increase in birth weight was associated with increased hours of sleep. These data allow estimates of associations with birth weight as precise as 80 g. Within the range of activities performed by pregnant women in the population studied, birth weight is unlikely to be associated with maternal physical activity.