Symptoms and disability vary considerably over time and among individuals with similar degrees of radiographic osteoarthritis. The results of recent clinical studies have reaffirmed that many patients prefer nonsteroidal anti-inflammatory drugs (NSAIDs) to analgesics while also demonstrating that many individuals who have been receiving NSAIDs can manage without them. Certain but as yet poorly characterized patients appear to respond more favorably to treatment with NSAIDs. Thus, both chronic unquestioning use of NSAIDs and denial of NSAIDs to individuals with osteoarthritis are extreme and inappropriate approaches to treatment. Concerns about the gastrointestinal toxicity of NSAIDs remain, although strategies to prevent this toxicity provide some reassurance and should be considered for use in individuals who are at particular risk. These strategies include the use of NSAIDs that are selective inhibitors of the prostaglandin endoperoxide synthase II isoenzyme and are potentially less ulcerogenic or the coprescription of prostaglandin analogues or other ulcer-healing drugs. The costs and benefits of such approaches have yet to be established. Cartilage metabolism is adversely affected by certain NSAIDs in vitro, but the relevance of these observations remains unclear. In our view, the judicious use of NSAIDs in the treatment of osteoarthritis is acceptably safe and indicated in individuals in whom pain cannot be managed by simple analgesics and nonpharmacologic approaches.