OBJECTIVE To assess the role of smoking on low birth weight (LBW). STUDY DESIGN From Massachusetts for 1998, 79,904 birth certificates were reviewed. Birth weight, gestational age, plurality and maternal race were analyzed in relation to the mother's smoking status during the pregnancy. The etiologic fraction (EF) was calculated for smoking and LBW for the group as a whole as well as for various subgroups. RESULTS A total of 11.7% of women acknowledged smoking during pregnancy. The overall LBW rate was 6.83%. The relative risk (RR) of LBW among smokers was 1.58. For all births the EF for smoking was 6.4% (95% CI: 5.4-7.3). For singleton pregnancies it was 10.9% (95% CI: 9.6-12.1) (14% for singleton whites and 7.2 for singleton blacks). At term, the EF of smoking on LBW was 13.4% (95% CI: 11.5-15.3), with an EF of 16.7% (95% CI: 14.5-18.7) for term singletons (21.4% among whites and 14.6% among blacks). Among very LBW infants, smoking accounted for 1.7% (95% CI:--0.5-3.8) of the outcome (5.8% among singletons). When stratifying for the effect of smoking, the rate of LBW was 6.38% among nonsmokers, 9.5% (RR 1.48, 1.38-1.61) among light smokers, 11.67% (RR 1.82, 1.63-2.05) among moderate smokers and 11.72% (RR 1.84, 1.33-2.54) among heavy smokers. Sixty percent of the overall population effect of smoking on LBW was in the category of light smokers. CONCLUSION The amount of LBW attributable to smoking was 6.4% in this sample. Among those who smoked, LBW was 58% more likely than among nonsmokers, and 60% of the overall population effect of smoking on LBW was noted among light smokers.