Objective: We previously reported that increased preoperative Beck Depression Inventory II (BDI-II) scores were associated with a 47% (95% CI 24%-64%) reduction in the rate of opioid cessation following surgery. We aimed to identify the underlying factors of the BDI-II (affective/cognitive vs somatic) associated with a decreased rate of opioid cessation after surgery. Methods: We conducted a secondary analysis of the data from a previously reported prospective, longitudinal, observational study of opioid use after five distinct surgical procedures (total hip replacement, total knee replacement, thoracotomy, mastectomy, and lumpectomy) in 107 patients. The primary endpoint was time to opioid cessation. After exploratory factor analysis of the BDI-II, mean summary scores were calculated for each identified factor. These scores were evaluated as predictors of time to opioid cessation using Cox proportional hazards regression. Results: The exploratory factor analysis produced three factors (self-loathing symptoms, motivational symptoms, emotional symptoms). All three factors were significant predictors in univariate analysis. Of the three identified factors of the BDI-II, only preoperative self-loathing symptoms (past failure, guilty feelings, self-dislike, self-criticalness, suicidal thoughts, worthlessness) independently predicted a significant decrease in opioid cessation rate after surgery in the multivariate analysis (HR 0.86, 95% CI 0.75-0.99, P value 0.037). Conclusions: Our results identify a set of negative cognitions predicting prolonged time to postoperative opioid cessation. Somatic symptoms captured by the BDI-II were not primarily responsible for the association between preoperative BDI-II scores and postoperative prolonged opioid use. © 2014 American Academy of Pain Medicine.