Background: We examined analgesic and anti-inflammatory medicine use by Australian veterans before and after the introduction of selective Cox-2 inhibitors. Methods: We studied cohorts of Gold Card-holding veterans using prescription data held by the Department of Veterans' Affairs for the period 1 July 1998 to 30 June 2004. Outcomes were volume dispensed, average daily quantity and cumulative incidence of use of paracetamol-containing and aspirin-containing medicines, non-selective and Cox-2-selective non-steroidal anti-inflammatory drugs (NSAIDs), tramadol and dextropropoxyphene. Results: Overall, we found high levels of use of analgesic and anti-inflammatory medicines, which increased by 43% over the study period. Use of paracetamol-containing medicines was overtaken by NSAIDs in 1999/2000, corresponding to the introduction of the Cox-2-selective agents. Between 12 and 17% of Cox-2-selective medicine recipients were supplied amounts indicative of continuous use in relatively high doses and 51% of veterans received at least one relatively Cox-2-selective medicine (celecoxib, rofecoxib, meloxicam, diclofenac) by the end of the study period. Dextropropoxyphene use declined during the study and tramadol use increased 10-fold. Conclusion: This study shows very high levels of Cox-2 inhibitor use during the 6-year period. Cox-2-selective agents were more likely to be taken continuously and at higher doses than non-selective NSAIDs. This is relevant in view of the cardiovascular toxicity of this group of medicines. The study shows the value of using unit record dispensing data to assess drug use patterns. Linking dispensing records to hospital separation and mortality data will further enhance our ability to monitor drug safety.