Cyclospora is a coccidian (previously referred to as cyanobacterium-like bodies) that has been implicated in cases of prolonged diarrhoea. The average duration of symptoms is more than three weeks, and no specific treatment has been shown to shorten the illness. A case report suggested that co-trimoxazole may be effective. Expatriate persons with gastrointestinal complaints and cyclospora detected on examination of faeces were recruited from two clinics in Kathmandu, Nepal, between May and August, 1994. Participants were assigned in a randomised, double-blinded manner to receive either cotrimoxazole (160 mg trimethoprim, 800 mg sulphamethoxazole) or placebo tablets twice daily for 7 days. Of 40 patients included in the study, 21 received cotrimoxazole and 19 placebo. There were no significant differences between these two groups in age, sex, time in Nepal, duration or severity of illness, or presence of other enteric pathogens. After 3 days, 71% of patients receiving co-trimoxazole still had cyclospora detected, compared with 100% of patients receiving placebo (p = 0.016). After 7 days, cyclospora was detected in 1 (6%) of 16 patients treated with co-trimoxazole who submitted stool specimens compared with 15 (88%) of 17 patients receiving placebo (p < 0.0001). Eradication of the organism was correlated with clinical improvement. There was no evidence of relapse of infection among treated patients followed for an additional 7 days. Treatment with co-trimoxazole for 7 days was effective in curing cyclospora infection among an expatriate population in Nepal.