OBJECTIVES: To examine the association between extended-release, once-daily morphine sulfate (Avinza®, A-MSER) and cognitive function (CF) while controlling for pain intensity (PI), pain-related emotions (PE), pain-related behaviors (PB), and benzodiazepine dose (BD). METHODS: Chronic non-malignant pain patients whose pain was not adequately controlled with short-acting narcotic analgesics and eligible for treatment with A-MSER were enrolled. Patients completed the following assessments at baseline and 4 weeks after stabilization of the A-MSER dose: PI, PE, PB and CF tests measuring short-term memory (digit span test – DST), information processing (paced auditory serial addition test – PASAT), and motor skills (digit symbol test – DSYT). Self-reports of average narcotic dose and BD in the week prior to the assessment were recorded at baseline and follow-up. Three structural equation models were developed; each CF test serving as the dependent variable and other assessments serving as predictor variables. RESULTS: Paired sample t-tests showed statistically significant (p <0.05) improvements in PI, PE, PB, and CF test scores. Chi-square fit statistics showed that the three models with DST (chi-square = 147.79, p = 0.76), DSYT (chi-square = 128.06, p = 0.5), and PASAT (chi-square = 160.39, p = 0.85) fit the data well. In each model, the association between narcotic dose and CF test scores was not statistically significant (p > 0.05) at baseline and follow-up. A statistically significant (p <0.05) inverse association between BD (1% dose increase) and DSYT scores (0.05% decline) and PASAT scores (0.06% decline) were observed at baseline; the direction of the association persisted but was not significant at follow-up. CONCLUSION: Narcotic therapy does not contribute to cognitive impairment in this sample of chronic pain patients, however, evidence of an inverse dose-dependent association between BD and tests assessing information processing and motor skills was observed.