The study sample was drawn from the RAND Adolescent Panel Study. The study followed a sample of West Coast youth from grade 7 to young adulthood, focusing on substance use and related behaviors over time. Data for the current analysis were collected in 1987 and 1989 at grades 10 and 12 (or the equivalent, since the sample includes school dropouts). Any illicit drug use in the past year was indicated by a response greater than zero on any item. Analyses included measures from several categories of drug- use antecedents: prior problem behavior, drug-related cognitions, social learning factors, drug availability, and indicators of mental health. Four measures of family bonding were included in the analyses: intact nuclear family, talks over problems with parents, parents do not respect the juvenile, and disruptive family events. The analysis examined the protective value of social bonds and whether certain social bonds have greater importance for some racial/ethnic groups. The researchers also explored the association of previous substance use with later "hard" drug use when social bonds and a broad range of other personal and environmental variables are statistically controlled. Bonds with family were inversely related to any use of illicit drugs other than marijuana; various forms of prior use were positively related to both any and frequent use; however, variables other than social bonds and prior use were equal or stronger predictors of both outcomes. Some differences were found across racial groups; African-Americans were less likely to use illicit drugs other than marijuana; Mexican-Americans were more affected by family factors than were other groups; and Asian- Americans were more affected by school failure. Implications for prevention are discussed. 7 tables and 67 references