Data from the 1980 National Natality Survey were used to investigate the effects of maternal smoking, height, weight, and educational attainment on the incidence of low birth weight among white non-Hispanic married mothers aged 20-34 years, live birth order 1-3. Using multiple logistic regression to control for the effects of infant's sex, live birth order, and maternal age, women with less than 12 years education have a low birth weight odds ratio of 2.38 and women with 12 years have a low birth weight odds ratio of 1.24 relative to women with 13 or more years. After further controlling for height, weight, and smoking, these odds ratios are reduced to 1.59 and 1.11, respectively. The majority of the reduction can be attributed to differences in smoking among the education groups. Women with low education are more likely to have smoked prior to pregnancy, more likely to smoke heavily, and less likely to stop smoking during pregnancy. The odds of low birth weight increases by 26% for every five additional cigarettes smoked per day. If all women in the study population stopped smoking during pregnancy, the incidence of low birth weight would be expected to decline by 35% for those with less than 12 years education, by 20% for those with 12 years, and by 11% for those with more than 12 years. The effects of the independent variables were also estimated for two components of low birth weight: below 2,000 grams and 2,000-2,499 grams. Although the nonresponse rate for the National Natality Survey was 20%, a comparison of mail and telephone respondents suggests that the impact of nonresponse on the estimates of the logistic coefficients is minimal.