OBJECTIVE Previous work found no effect on birthweight of alcohol and caffeine consumption in non-smokers but such an effect was found in smokers. This report investigates further the effects on birthweight of alcohol and caffeine at three stages of pregnancy in smoking women. METHODS This was a prospective population study. METHODS District general hospital in inner London. METHODS Out of 1309 women who completed all pregnancy interviews, 895 were excluded because they did not smoke, leaving a sample of 414 smokers. RESULTS Number and brand of cigarettes smoked, and quantity of alcohol and caffeine consumed were obtained by interview at booking, 28, and 36 weeks gestation. Birthweight was corrected for gestation and adjusted for maternal height, sex of infant and parity. The effect on birthweight of alcohol consumption was not explained by the amount smoked in terms of quantity and yield. Similarly the effect of caffeine was independent of smoking. When alcohol, caffeine, and smoking were analysed together, alcohol and caffeine were both associated with reductions in birthweight. Alcohol was associated with a reduction of up to 8% after adjusting for tobacco and caffeine intake, and caffeine was associated with a reduction of up to 6.5% after adjusting for tobacco and alcohol intake. Women who at booking were heavy smokers (greater than or equal to 13 cigarettes/day or greater than or equal to 15 mg carbon monoxide/cigarette), heavy drinkers (greater than or equal to 100 g/week alcohol), and had high caffeine intake (greater than or equal to 2801 mg/week) had a predicted reduction in mean birthweight of 18% (95% CI 11% to 24%). CONCLUSIONS It is well known that women who smoke in pregnancy have smaller babies than non-smokers. Our study suggests that if these women also drink alcohol and high quantities of caffeine then the risk of poor fetal growth is increased even further.