Selective cyclooxygenase-2 inhibitors have been marketed as alternatives of conventional, non-steroidal anti-inflammatory drugs with the purpose of reducing/eliminating the risk of ulcer complications. Unexpectedly, randomized-controlled trials revealed that long-term use of coxibs, such as rofecoxib, significantly increased the risk of myocardial infarction and stroke, while the use of valdecoxib was associated with potentially life-threatening skin reactions. Subsequently, rofecoxib and valdecoxib were withdrawn from the market. Although more strict precautions for other coxibs, such as celecoxib, etoricoxib, lumiracoxib and parecoxib, may be accepted/recommended by regulatory agencies, a critical review of published data suggests that their use may not be justified - even in high-risk patients - taking benefits, costs and risks into consideration. Clinicians should, therefore, never prescribe coxibs to patients with cardiovascular risk factors, and should only reluctantly prescribe coxibs to patients with a history of ulcer disease or dyspepsia to overcome persistent pain due to, e.g. rheumatoid arthritis or osteoarthritis. Instead, they should consider using conventional non-steroidal anti-inflammatory drugs in combination with a proton pump inhibitor or a prostaglandin analogue, especially for patients with increased cardiovascular risks, i.e. established ischaemic heart disease, cerebrovascular disease and/or peripheral arterial disease, or alternatively acetaminophen. An evidence-based algorithm for treatment of a chronic arthritis patient with one or more gastrointestinal risk factors is presented.