Cigarette smoking among adolescents was predicted by age, race, involvement in activities with peers and family, peer smoking, family's smoking, and peers' acceptance of smoking. A random sample of 1,826 South Carolina public school students served as subjects. Multiple regression analysis indicated that 43% of the variance in reported cigarette smoking could be explained by these variables. Older, white subjects and subjects with friends and family members who smoked were more likely to smoke themselves. Adolescents who reported low family involvement and high peer involvement were more likely to smoke. High peer involvement and peers' approval of smoking were also associated with greater reported smoking.