Chronic pain in the elderly is frequently a result of arthritic disorders, particularly osteoarthritis. The cyclo-oxygenase (COX)-2 inhibitors are as effective as standard NSAIDs for the relief of pain and for improving function in elderly patients with osteoarthritis and rheumatoid arthritis. COX-2 inhibitors increase the risk of serious gastroduodenal adverse reactions but there is evidence that they carry a lower risk for these adverse effects than standard NSAIDs, except when there is concurrent aspirin use. Since gastroduodenal disorders are the most frequently reported serious adverse effects of NSAIDs and these disorders occur more frequently in the elderly, COX-2 inhibitors offer an alternative to standard NSAIDs in this age group. However, they are not appropriate for many patients with cardiovascular and renal disease. The adverse reaction profile of the COX-2 inhibitors has confirmed the role of the COX-2 enzyme in renal function, salt and water homeostasis and the vascular endothelium. Thus, like standard NSAIDs, COX-2 inhibitors can cause renal failure, hypertension and exacerbation of cardiac failure. Of note is that these disorders are dose related. Thus, there are good reasons to avoid high doses of COX-2 inhibitors in the elderly. Clinical trials indicate that daily doses of rofecoxib 12.5 mg, celecoxib 100-200 mg, valdecoxib 10mg and etoricoxib 60 mg are the minimum effective doses of these agents. Data from the New Zealand Intensive Medicines Monitoring Programme indicate that celecoxib 200 mg/day and rofecoxib 25 mg/day are/were the most commonly prescribed doses and that 6% of patients had taken rofecoxib 50 mg/day for longer than recommended. Recent research indicates that COX-2 inhibitors have a thrombotic potential, especially in high doses and when use is prolonged, and this further limits the extent to which they can be used in the elderly. Important interactions with COX-2 inhibitors in the elderly include those with warfarin, which can result in loss of control of anticoagulation, and those with ACE inhibitors, angiotensin II type 1 receptor antagonists and diuretics, which can result in loss of control of blood pressure and cardiac failure and, in hypovolaemic conditions, renal failure. The clinical significance of an interaction between celecoxib and aspirin to reduce the antiplatelet effect of the latter drug is unknown. Preliminary information from spontaneous reporting systems indicates that there may be differences in the risk of cardiac failure and hypertension between standard NSAIDs and COX-2 inhibitors and between rofecoxib and celecoxib. More formal studies using equivalent doses are needed to test this observation. Use of COX-2 inhibitors may be considered in the elderly to reduce the risk of gastroduodenal complications associated with standard NSAIDs but only when consideration has first been given to use of less toxic medicines as alternatives or supplements, the appropriate dose of the COX-2 inhibitor or standard NSAID, the presence and possible impact of co-morbidities, and the implications of taking COX-2 inhibitors with any concomitant medications. Equally important is regular monitoring of the patient taking a COX-2 inhibitor for efficacy and adverse effects, and ensuring that the patient has a continuing need to keep taking the drug. Close attention also needs to be paid to intercurrent illnesses and new prescriptions that may reduce the safety of the COX-2 inhibitor. A standard NSAID plus a proton pump inhibitor may be equally effective as a COX-2 inhibitor in reducing the risk of gastroduodenal toxicity and if used the same prescribing advice applies. Current knowledge concerning the thrombotic potential of COX-2 inhibitors suggests that this combination, if tolerated, may be preferable to a COX-2 inhibitor, particularly where prolonged use is required. This knowledge also indicates that for patients with or at high risk of ischaemic heart disease or stroke, COX-2 inhibitors are contraindicated.