Data from a nationwide cross-sectional survey of the smoking status among junior and senior high school students in Japan conducted in 1990, were utilized to provide sociodemographic correlates of the use of cigarettes. The study was based on data from 57, 189 students. The analyses examined the relationship between smoking status and 9 explanatory variables, using multiple logistic regression for both males and females. The results of logistic regression analyses revealed the following main effects: 1) Smoking by a friend had the strongest relationship to smoking status of students for both sexes. The relative risks for having a smoking experience if a friend smoked were 3.45 for junior high school boys, 5.24 for junior high school girls, 3.65 for senior high school boys and 5.54 for senior high school girls and the risks for current smoking if a friend smoked were 8.54, 15.28, 6.99 and 12.47, respectively. 2) Among both boys and girls, smoking experience and current smoking was significantly more likely if a brother or sister, or mother smoked. Awareness of the harm of smoking was also an important correlate in both sexes. Whether the students enjoyed their school life was significantly related to current smoking status for females, but not for males. 3) The relative risks for current smoking if their mothers smoked were 1.60 for junior high school boys, 1.78 for junior high school girls, 1.61 for senior high school boys and 1.61 for senior high school girls. Fathers' smoking was not selected in this stepwise multiple logistic regression model. Therefore, mothers' smoking appears to be a stronger correlate of smoking status among students than fathers' smoking. 4) The relative risks for smoking experience and current smoking if there were smokers in the family, were higher among junior high school students than senior high school students.