This study tests a broad array of determinants of utility for developing smoking preventive interventions using a population-based cohort of early adolescents. Multivariable logistic regressions using never-smokers at baseline (N=1137; age 11-14) showed a model of distal determinants was more predictive of initiation within the approximate 10 month follow up period than one of proximal determinants. When all determinants were simultaneously considered, lesser academic achievement and fewer environmental impediments to smoking most strongly predicted initiation. The findings are consistent with some current smoking prevention programs, however such programs may be further potent by using theory-based social development approaches and through reducing tobacco availability or social contexts where youth can smoke without another adult knowing.