Background: Prior authorization (PA) is a poorly studied but commonly employed policy used by health care payers to manage the rising costs of pharmacy benefits. Objective: The aim of this study was to evaluate the intended and unintended effects of a PA policy for celecoxib on pharmacy and medical-service utilization in a Medicaid managed-care organization. Methods: This was a retrospective, interrupted time-series analysis of 22 monthly health-related utilization rates from January 1, 1999, to October 31, 2000. All Medicaid claims for CareOregon (a managed-care organization) and a fee-for-service program were reviewed. A model was constructed to evaluate changes in utilization of therapeutically related drug classes (eg, conventional nonsteroidal anti-inflammatory drugs [NSAIDs], gastrointestinal agents), office and emergency-department encounters, and hospitalizations before and after the PA policy was implemented on November 16, 1999. A secondary analysis evaluated these changes among a sample of prior NSAID users. Results: After the PA policy was implemented, use of celecoxib was immediately reduced from 1.07 to 0.53 days' supply per person-year (58.9%; 95% CI, 50.0%–67.9%). The monthly rate of increase was also reduced ( P < 0.001). Utilization changes were not observed in other drug classes. Similar changes were observed in the secondary analysis. An 18% (95% CI, 2.2%–33.9%) nonsignificant increase in emergency-department visits was observed in the entire sample after the PA policy was implemented. However, a similar change was not observed in the secondary analysis of prior NSAID users. No other changes in medical service encounters were noted after the PA policy was activated. Conclusions: This observational study found that celecoxib use was substantially reduced after the implementation of a PA policy. No important changes in use of other drug classes were detected. The overall increase in emergency-department visits-although not observed among previous NSAID users-should be explored on the individual level.