Non-selective non-steroidal anti-inflammatory drugs (NSAIDs) have been associated with adverse gastrointestinal (GI) events. Rofecoxib was introduced as a drug causing fewer GI symptoms than non-specific NSAIDs. Merck voluntarily removed it from the Canadian in September 2004 due to a possible association with cardiovascular events. No studies have examined the impact of its removal on utilization of other NSAIDs in Canada. OBJECTIVES: We examined the utilization (i.e., numbers of claims) of NSAIDs, proton-pump inhibitors (PPIs) and salicylic acid derivatives (SADs) in two Canadian provinces one year prior to and after rofecoxib's market withdrawal. METHODS: We obtained monthly drug claims (Oct/03-Nov/05) from Brogan Inc. We used interrupted time-series design by segmented regression to examine utilization changes of other coxibs, COX-1 NSAIDs, PPIs and SADs after rofecoxib's removal. Series were transformed using ARIMA and other appropriate models to account for autocorrelation, non-stationarity and heteroscedasticity. RESULTS: All time series were fit with either an ARIMA or a log linear model. Each model showed excellent fit (R2=0.79-0.98). Segmented regression analysis revealed a marked reduction in total claims for all NSAIDs between September 2004 and January 2005 (-141,524 claims/month; CI95 = -188,889 to -94,159; p<0.001). The month-to-month variation before and after September 2004 was not significantly different than zero. Monthly coxib claims decreased after rofecoxib's withdrawal (-156,883 claims/month; CI95 = -184,393 to -129,373; p<0.001). Non-selective COX-1 inhibitors exhibited the opposite trend; monthly claims increased 32% to 414,813 at study end. No shift in utilization of PPIs or SADs was observed. CONCLUSION: The removal of rofecoxib from the Canadian market, using two provinces as a proxy, was associated with an immediate decrease followed by a slow increase in NSAID claims. Many patients discontinued NSAIDs, while some switched to non-selective NSAIDs as an alternative.