Objectives The emergence of HIV drug resistance is a crucial issue in A frica, where second‐line antiretroviral therapy ( ART ) is limited, expensive and complex. We assessed the association between adherence patterns and resistance emergence over time, using an adherence measure that distinguishes low adherence from treatment interruptions, in rural C ameroon. Methods We performed a cohort study among patients receiving nonnucleoside reverse transcriptase inhibitor ( NNRTI )‐based ART in nine district hospitals, using data from the S tratall trial (2006−2010). Genotypic mutations associated with antiretroviral drug resistance were assessed when 6‐monthly HIV viral loads were > 5000 HIV ‐1 RNA copies/mL. ART adherence data were collected using face‐to‐face questionnaires. Combined indicators of early (1−3 months) and late (6 months to t − 1; t is the time point when the resistance had been detected) adherence were constructed. Multivariate logistic regression and Cox models were used to assess the association between adherence patterns and early (at 6 months) and late (after 6 months) resistance emergence, respectively. Results Among 456 participants (71% women; median age 37 years), 45 developed HIV drug resistance (18 early and 27 late). Early low adherence (< 80%) and treatment interruptions (> 2 days) were associated with early resistance [adjusted odds ratio (95% confidence interval) 8.51 (1.30–55.61) and 5.25 (1.45–18.95), respectively]. Early treatment interruptions were also associated with late resistance [adjusted hazard ratio (95% confidence interval) 3.72 (1.27–10.92)]. Conclusions The emergence of HIV drug resistance on first‐line NNRTI ‐based regimens was associated with different patterns of adherence over time. Ensuring optimal early adherence through specific interventions, adequate management of drug stocks, and viral load monitoring is a clinical and public health priority in A frica.