The relative toxicity of phenylbutazone, flunixin meglumine, and ketoprofen was studied in healthy adult horses. Sixteen horses were randomly assigned to receive 10 ml of physiologic saline solution, or ketoprofen (2.2 mg/kg of body weight), flunixin meglumine (1.1 mg/kg), or phenylbutazone (4.4 mg/kg) IV, every 8 hours, for 12 days. Results of CBC, serum biochemical analyses, and fecal occult blood tests were monitored. On day 13, all horses were euthanatized and complete necropsy examinations were performed. Mean CBC values remained within normal limits for all groups. Phenylbutazone-treated horses had a significant (P < 0.05) decrease in serum total protein and albumin concentrations. Mean values of all other serum biochemical assays were not different from those of the saline-treated group. Results of all fecal occult blood tests were negative. At necropsy, the glandular portion of the stomach was the area of the gastrointestinal tract most severely affected by phenylbutazone, flunixin meglumine, and ketoprofen. In the phenylbutazone-treated group, but not in the other groups, edema of the small intestine and erosions and ulcers of the large colon were observed. None of the horses treated with saline solution had lesions in the glandular portion of the stomach or in the intestine. Four horses (1/5 and 3/3 in the flunixin- and phenylbutazone-treated groups, respectively) developed renal crest necrosis. Horses in the saline- and ketoprofen-treated groups did not develop renal lesions.(ABSTRACT TRUNCATED AT 250 WORDS)