Objective: Adults with serious mental illness (SMI) have high tobacco use disorder (TUD) rates and underutilization of first-line TUD pharmacotherapy. In a randomized trial, a community health worker (CHW) plus primary care provider (PCP) education (PE) intervention improved tobacco abstinence in this population at two years, partly through increasing TUD pharmacotherapy initiation. The current study determined how engagement between participants and CHWs was associated with these outcomes. Methods: This was a secondary, mixed-methods analysis of 196 participants in the PE+CHW arm of the RCT. Effects of CHW visit number and duration (minutes), CHW co-led smoking cessation group sessions attended, and CHW-attended PCP visit number on TUD pharmacotherapy initiation and tobacco abstinence were modeled using logistic regression. Interviews with 12 CHWs, 16 participants, and 17 PCPs were analyzed thematically. Results: Year-two tobacco abstinence was associated with one standard deviation increase in CHW visit number (OR=1.85, z-score [1.29, 2.66]) and duration (OR=1.85, [1.33, 2.58]) and number of groups attended (OR=1.51, [1.00, 2.28]); effects on TUD pharmacotherapy initiation were similar. 1-3 CHW visits per month over two years were optimal for achieving abstinence. Interviews identified engagement facilitators, including CHWs providing goal accountability, skills reinforcement, and assistance in overcoming barriers to treatment access and adherence. Robust training and supervision facilitated CHW effectiveness. Barriers included PCPs' and care teams' limited understanding of the CHW role. Conclusions: Greater CHW engagement within feasible dose ranges was associated with increased tobacco abstinence in adults with SMI. Implementation of CHW interventions may benefit from promoting CHW training and integration within clinical teams.