Etodolac, a new anti-inflammatory analgesic drug found to be effective in treating arthritis in a dose range of 100 to 300 mg bid, has been shown to induce significantly less gastrointestinal microbleeding in normal men than several other NSAIDs. In this study, the effect on gastrointestinal blood loss of high-dose etodolac, 300 and 500 mg bid, versus piroxicam at its normal therapeutic dose of 20 mg qd, was investigated by the 51Cr method in 23 men with osteo- or rheumatoid arthritis. Placebo periods preceded and followed 28 days of active drug treatment. Blood and stool analyses were performed by an analyst not aware of drug assignment or study design. Patients receiving piroxicam, but not those receiving either dose of etodolac, had a significantly higher mean level of fecal blood loss in the active treatment phase compared with the pretreatment placebo level (p less than 0.01). Further, microbleeding was significantly greater for the piroxicam group during treatment than for either of the etodolac groups (p less than 0.01). There were no significant differences in fecal blood loss between the two groups receiving etodolac compared with pretreatment. Even at doses two to three times those found effective in the treatment of arthritis, etodolac produces no increase in fecal blood loss, in contrast to blood loss seen with the recommended dose of piroxicam. Fecal blood loss in osteoarthritic patients, not receiving an NSAID, was similar to normal subjects in previous studies.