Aims: To investigate the relationship between acute coronary syndrome (ACS) and ingested doses of selective cyclooxygenase-2 (COX-2) inhibitors and other nonsteroidal anti-inflammatory drugs (NSAIDs). Methods: Case-control study, commenced August 2003. Cases were patients admitted to hospital with ACS (myocardial infarction/unstable angina). Controls were hospital patients admitted for reasons other than acute vascular ischaemia or conditions that are believed to be complications of treatment with COX-2 inhibitors or NSAIDs. Structured interviews were undertaken within 7 days of admission, collecting information on cardiovascular events and risk factors and all ingested drugs, including the doses of COX-2 inhibitors and NSAID consumed in the previous week and month. Results: An interim analysis of the data was conducted in late 2004 to inform a review of the COX-2 inhibitors by the Australian drug regulatory agency. Between August 2003 and October 2004, we recruited 328 ACS cases and 478 controls. With non-use of COX-2 inhibitors or NSAIDs as the reference the adjusted odds ratios (OR) for ACS were: celecoxib 1.11 (95% confidence interval 0.59, 2.11), rofecoxib 0.63 (0.31, 1.28) and other NSAIDs 0.67 (0.41, 1.09). Among control subjects, median daily ingested doses of celecoxib and rofecoxib were 200 mg and 13.4 mg, respectively. Using these to stratify risk, adjusted ORs for ACS were: 'low' dose (< median) 0.44 (0.19, 1.03); 'high' dose (≥ median) 1.22 (0.67, 2.21). A test for interaction across doses was statistically significant, OR 2.8 (1.0, 7.7), suggesting that at low doses, COX-2 inhibitors may be protective, becoming risk-inducing only at higher doses. Conclusion: The possibility that the gradient of cardiovascular risk with COX-2 inhibitors runs from protective to risk-inducing has biological plausibility and merits further investigation.