Consistent with previous research, males were almost three times as likely as females to report drinking before age 13 and almost twice as likely as females to report drinking before ages 13 to 15. Later-born respondents reported earlier drinking than first-borns, perhaps reflecting their greater opportunity to imitate the behavior of older siblings. African-American respondents were significantly less likely to report early drinking than their non-Hispanic, non-African-American peers. Drinking also began earlier for respondents who were not living with both parents at age 14, had ever been charged with an illegal act, had a family history of alcohol problems, had lower academic aptitude, and attended religious programs less frequently than other youth. Analyses confirmed that ADO measures obtained in 1982 or 1983 predicted alcohol disorders measured in 1994, confirming the findings of Grant et al. (2001). The source of the study data, the National Longitudinal Survey of Youth, is based on a nationally representative sample of youth ages 14 to 22 in 1979. The panel, which consisted of 12,686 respondents in 1979, has been followed annually through 1994 and every other year thereafter. Assessment of the predictors of early ADO was based on 8,165 respondents who supplied complete data on ADO and its possible precursors. Analyses of the mediators of the relationship between early ADO and later alcohol misuse were based on 5,643 respondents who had complete data on ADO, contextual variables, and alcohol disorders in 1989 and 1994. 3 tables and 24 references