Whipple's disease is a multisystemic disorder in which almost all organ systems can be invaded by rod-shaped bacteria. Without extended antimicrobial therapy, its course is lethal. Empirically, treatment consists of tetracyclines given for one to two years. Trimethoprim-sulfamethoxazole, a compound that crosses the blood-brain barrier, has been suggested as an alternative when patients were observed with progressive cerebral involvement. There has never been a formal evaluation of the selection of antibiotics for the treatment of Whipple's disease. In the present nonrandomized, partially retrospective study, we compared the result of two treatment regimens in 30 patients, all examined personally. Twenty-two patients were treated with tetracycline and eight patients with trimethoprim-sulfamethoxazole. In five patients, therapy with tetracycline was changed to another antimicrobial agent because of treatment failure or drug intolerance. The main treatment measure was disappearance of the clinical symptoms such as weight loss, arthritis, malabsorption, fever, edema, central nervous system manifestations, lymphadenopathy, and congestive heart failure. Drug intolerance requiring a change of medication was also considered a treatment failure. We found that trimethoprim-sulfamethoxazole induced complete clinical remission in 12 of 13 treatment cycles, tetracycline in 13 of 22 treatment cycles (P < 0.05; mean difference 33%; 95% confidence interval 8% to 58%). Trimethoprim-sulfamethoxazole was also more efficacious than tetracycline in the treatment of cerebral Whipple's disease. However, trimethoprim-sulfamethoxazole did not prevent cerebral manifestations in all cases. The only deaths due to Whipple's disease occurred in patients with cerebral involvement. It is concluded that treatment with trimethoprim-sulfamethoxazole was significantly superior to that with tetracycline in inducing clinical remission of Whipple's disease.(ABSTRACT TRUNCATED AT 250 WORDS)