AIMS: This study examines 12-Step attendance trajectories over 7 years, factors associated with the trajectories, and relationships between the trajectories and long-term substance use outcomes among adolescents entering out-patient substance use treatment in a private, non-profit integrated managed-care health plan. DESIGN: Longitudinal observational study. SETTING: Four Kaiser Permanente Northern California substance use treatment programs. PARTICIPANTS: A total of 391 adolescents entering treatment between 2000 and 2002 who completed at least one follow-up interview in year 1, and at least one during years 3-7, after treatment entry. MEASUREMENTS: Alcohol and drug use, 12-Step meeting attendance and activity involvement and post-treatment medical service utilization. FINDINGS: Semiparametric group-based modeling identified three distinct 12-Step attendance trajectory groups over 7 years: low/no attendance (60%), early but not continued (26%) and continued (14%). There were lower proportions of males and of adolescents with prior substance use treatment experience in the low/no attendance group (P = 0.019 and P = 0.003, respectively). In addition, those in the low/no attendance group had lower perception on circumstances, motivation and readiness for treatment at baseline (P = 0.023). Multivariate logistic generalized estimating equation analyses found that those in the continued group were more likely to be abstinent from both alcohol and drugs during follow-ups than those in the low/no attendance group [odds ratio (OR) = 2.40, P = 0.003 and OR = 1.96, P = 0.026, respectively]. However, no differences in long-term outcomes were found between those in the other two groups. CONCLUSION: Robust connection with 12-Step groups appears to be associated with better long-term outcomes among adolescents with substance use disorders.