Of the nine significant predictors of early ADO in the final main effects model, three appeared before 6 years old. The earliest predictor was mother's ADO (drinking by age 14 or earlier). This variable was amplified by family history of alcohol problems. Being female was also a significant predictor of childhood ADO. The last predictor was the Behavior Problems Index (BPI) measure of "headstrong" collected by age 6. The other main-effect predictors were all collected at the baseline year. Age itself predicted ADO, with the developmental period between ages 11 and 14 being critical for ADO. The incidence of ADO accelerated during these years. Two related parental variables each independently predicted early ADO, i.e., living with mother only versus with both parents and no adult being home when school lets out. Children with more problem behaviors reported earlier ADO. Those with three or more problems had their risk more than doubled compared to children with no such problems. Feeling more often lonely was associated with delayed ADO. The study also found that early drinking should be a concern not only for boys but also for girls and that the gender risk of ADO was complicated by other contextual factors. One research recommendation is that long-term follow-up studies be conducted to compare adult respondents with experimentally delayed ADO to a control group, so as to assess their relative risk of alcohol-related problems. Such evidence is absent for the causal influence of ADO on later alcohol abuse. The sample consisted of children of the women respondents in the National Longitudinal Survey of Youth who were 11 or 12 years old and still not drinking in the baseline interviews and who were reinterviewed 2 years later (n=1,951). 2 tables, 38 references, and sample characteristics for variables not found significantly associated with drinking onset