Identifying the critical aspects of family relationships that are associated with adolescents not smoking may provide valuable information for designing successful parent-child interventions. The purpose of this study was to identify which aspects of family life are predictive of adolescent initial experimentation and daily cigarette use, using a wide-ranging set of indicators of family relationships. The data source was the National Youth Survey, which contained a sample of youth aged 11 to 17. The adolescents were assessed periodically over a 7-year period. Parent interviews were obtained at the initial screening. Nineteen indicators of family relationships were used based on parent and child interviews. Several indicators, such as first time cigarette smoking and a nine-point scale of intensity of experimentation with cigarettes, were used. Logistic and multiple regression analyses that controlled for ethnicity, gender, and socioeconomic status found that family relationships were only predictive for subjects under the age of 18. The results suggest that there are many different aspects of family life that predict smoking, and many aspects of these relationships are more predictive of daily smoking than they are of initial smoking. Even after initial experimentation with cigarettes, positive family relationships may reduce the chance of an adolescent becoming a regular smoker. Psychosocial factors that predict experimentation may play a greater or lesser role in the development of daily smoking. Several new aspects of family life were identified as predictors of smoking, including perceived parental disapproval, dishonesty toward parents, frequency of participation in family activities, and method of discipline. 3 tables, 22 references