The association of early alliance to treatment attendance and longitudinal outcomes were examined in 356 adolescents participating in a randomized clinical trial targeting cannabis use. Both patient and therapist views of alliance were examined, and outcomes were evaluated over 12 months after numerous other sources of variance were controlled. Patient-rated alliance predicted a reduction in cannabis use at three and six months and a reduction in substance-related problem behaviors at six months. Therapist-rated alliance did not predict outcomes. Neither patient nor therapist alliance ratings were associated with attendance. The findings support the important and often overlooked role that alliance can play in treating substance abusing, often delinquent, adolescents.