Nonsteroidal anti-inflammatory drugs (NSAIDs) inhibit the enzyme cyclooxygenase (COX), which forms prostaglandins involved in pain and inflammation. COX inhibitors have analgesic and antiinflammatory effects, but also increase risks for gastrointestinal ulcers, bleeding, and renal and cardiovascular adverse events. Identification of two isoforms of COX, COX-1 and COX-2, led to the development of selective COX-2 inhibitors, which were launched as having fewer gastrointestinal side effects since gastroprotective prostaglandins produced via COX-1 are spared. The balance between COX-1 mediated prothrombotic thromboxane and COX-2 mediated antithrombotic prostacyclin is important for thrombotic risk. An increased risk of suffering myocardial infarction and death with COX-2 inhibitor treatment is well established from clinical trials and observational research. Rofecoxib (Vioxx) was withdrawn from the market for this reason, but the equally COX-2 selective etoricoxib has replaced it in Europe but not in the United States. The "traditional" NSAID diclofenac is as COX-2 selective as celecoxib and increases cardiovascular risk dose dependently. COX inhibitor dosages should be lower in osteoarthritis than in rheumatoid arthritis. Randomized trials comparing COX-2 inhibitors with NSAIDs have exaggerated their gastrointestinal benefits by using maximal NSAID doses regardless of indication, and/or hidden the cardiovascular risk by comparing with COX-2 selective diclofenac instead of lowdose ibuprofen or naproxen. Observational studies show increased cardiovascular risks within weeks of treatment with COX-2 inhibitors and high doses of NSAIDs other than naproxen, which is the safest alternative. COX inhibitors are symptomatic drugs that should be used intermittently at the lowest effective dosage, especially among individuals with an increased cardiovascular risk.