The efficacy and safety of etodolac and piroxicam in patients with active rheumatoid arthritis were compared. A 12 week, double blind, parallel group study was conducted at 28 centers in patients 18 to 75 yrs old, randomized to receive etodolac or piroxicam. Primary efficacy criteria were investigators' and patients' global assessments and numbers of painful and swollen joints. Secondary criteria were duration of morning stiffness, grip strength, time to walk 50 feet. Westergren erythrocyte sedimentation rate (ESR), pain intensity, painful and swollen joint scores, and articular index. Of 426 patients enrolled, 140 received etodolac 200 mg bid (E200), 147 received etodolac 300 mg bid (E300), and 139 received piroxicam 20 mg qd (P20). Efficacy analyses included data from 361 patients. No significant differences occurred between the E300 and piroxicam groups in change from baseline for the primary efficacy variables. All treatments produced significant (p < 0.01) improvement from baseline in all efficacy variables, except that only E300 produced a significant decrease from baseline in ESR. No significant differences occurred in the incidence of any specific adverse event. Six patients given E200, 7 given E300, and 10 given piroxicam withdrew because of adverse reactions. The incidence of patients with hemoglobin and hematocrit results below the normal range was significantly higher for piroxicam than for either etodolac regimen. Three patients receiving piroxicam had gastrointestinal ulcers. Thus, E300 and P20 provided comparable efficacy. E200 was less effective for some variables early in the study.