This study explored predictors of community-based providers' adherence to MATCH, a modular cognitive behavioral therapy for children and adolescents. Provider-reported adherence to MATCH was measured using three increasingly strict criteria: (1) "session content" (whether the session covered MATCH content consistent with the client's target problem), (2) "session content and sequencing" (whether the session covered MATCH content in the expected sequence for the client's target problem), and (3) "session content, sequencing, and participant" (whether the session covered "MATCH content" in the expected sequence and with the expected participant(s) for the client's target problem). Session, client, provider, and organizational predictors of adherence to MATCH were assessed using multilevel modeling. Results revealed that nearly all providers delivered MATCH content that corresponded to the target problem, but only one-third of providers delivered MATCH "content" in the expected "sequence" and with the expected "participant" for the client's target problem. This difference underscores the need for nuanced adherence measurement to capture important implementation information that broad operationalizations of adherence miss. Regardless of the criteria used providers were most adherent to MATCH during sessions when clients presented with interfering comorbid mental health symptoms. This suggests that the design of MATCH, which offers flexibility and structured guidance to address comorbid mental health problems, may allow providers to personalize treatment to address interfering comorbidity symptoms while remaining adherent to evidence-based practices. Additional guidance for providers on managing other types of session interference (e.g., unexpected events) may improve treatment integrity in community settings.