Several currently available nonsteroidal anti-inflammatory drugs (NSAIDs) have been evaluated for their relative selectivity in inhibiting the two cyclooxygenase (COX) isozymes, COX-1 and COX-2. Arguments have been made that more selective inhibitors of COX-2 will be safer than less selective ones. Rankings of the COX-2/COX-1 inhibition ratios of various NSAIDs as they relate to the agents' toxicities have been used as evidence that COX-2 selectivity is an important factor in the upper gastrointestinal (GI) safety of some NSAIDs. Unfortunately, none of these claims has been supported by endoscopy studies in treated patients. Since all NSAIDs inhibit COX-1, they all cause upper GI mucosal damage. What is needed are specific COX-2 inhibitors that do not inhibit COX-1. Such agents are currently under development. Ongoing clinical trials will determine the potential role for specific COX-2 inhibitors in the treatment of arthritis and pain. If specific COX-2 inhibitors are shown to be both safe and effective, the treatment of rheumatic diseases will be revolutionized.