This article describes the patterns and effects of maternal snuff use, cigarette smoking and exposure to environmental tobacco smoke during pregnancy on birthweight and gestational age, in women living in Johannesburg and Soweto in 1990. A cohort of 1593 women with singleton live births provided information about their own and household members' usage of tobacco products during pregnancy. The women completed a questionnaire while attending antenatal services. Data on gestational age and birthweight were obtained from birth records. Women who smoked cigarettes or used snuff during pregnancy accounted for 6.1% and 7.5% of the study population respectively. The mean birthweight of non-tobacco users was 3148 g [95% CI 3123, 3173] and that of the smokers 2982 g [95% CI 2875, 3090], resulting in a significantly lower mean birthweight of 165 g for babies of smoking mothers (P = 0.005). In contrast, women using snuff gave birth to infants with a mean birthweight of 3118 g [95% CI 3043, 3192], which is a non-significant (P = 0.52) decrease (29.4 g) in their infants' birthweights compared with those not using tobacco. A linear regression analysis identified short gestational age, female infant, a mother without hypertension during pregnancy, coloured (mixed racial ancestry), and Asian infants compared with black infants, lower parity, less than 12 years of education and smoking cigarettes as significant predictors of low birthweight, while the use of snuff during pregnancy was not associated with low birthweight. The snuff users, however, had a significant shorter gestational age than the other two groups of women. The birthweight reduction adjusted for possible confounders was 137 g [95% CI 26.6, 247.3 (P = 0.015)] for cigarette smokers and 17.1 g [95% CI -69.5, -102.7, P = 0.69] for snuff users respectively, compared with the birthweight of non-tobacco users. Among women who did not smoke cigarettes or use snuff, exposure to environmental tobacco smoke did not result in significant effects on the birthweight of their infants. In conclusion, infants of cigarette smokers had significantly lower birthweights than those of non-tobacco users or snuff users who are exposed to nicotine during pregnancy. Passive smoking did not affect birthweight significantly in this population.