The 63 schools were stratified into 5 parental income levels, and 4 schools were randomly selected from each of the 5 income levels. Initial work was done in 1975 using 2,583 children in grades 4 through 6, and subsequent work was carried out in 1978 using 2,372 children in grades 6 through 8. Results were available for 1,682 children who responded on both occasions. An 83-item questionnaire was administered to study participants. Results of multivariate analysis showed current smoking by young people was related to age, socioeconomic status, school achievement, attitudes, knowledge, and parental and peer smoking. There was no association between sex and smoking. Exposure to a smoking awareness curriculum did not appear to have any direct effect on the smoking behavior of children. 27 references and 6 tables