OBJECTIVE: To evaluate the effect of highly active anti-retroviral therapy (HAART) and cotrimoxazole prophylaxis on morbidity after HAART eligibility. METHODS: Between 1999 and 2006, we collected morbidity data from a community-based cohort of HAART-eligible patients, comparing patients initiating HAART and those non-HAART. Patients aged 15 years or older visited the clinic every 6 months and when ill. Baseline data on patients' characteristics, WHO stage, haemoglobin and CD4+ T-cell counts, along with follow-up data on morbidity (new, recurrent and drug-related), were collected for the first year after initiating HAART or becoming HAART-eligible. We estimated the overall effect of HAART on morbidity; adjusted for the effect of cotrimoxazole prophylaxis by Mantel-Haenszel methods. A negative binomial regression model was used to assess rate ratios (RR) after adjustment for other confounders, including cotrimoxazole. RESULTS: A total of 219 HAART patients (median age 37 years; 73% women; 82% using cotrimoxazole prophylaxis, median haemoglobin 11.7 g/dl and median CD4+ 131 cells/microl) experienced 94 events in 127 person-years. 616 non-HAART patients (median age 33 years; 70% women; 26% using cotrimoxazole prophylaxis, median haemoglobin 11.2 g/dl and median CD4+ 130 cells/microl) experienced 862 events in 474 person-years. The overall morbidity during the first year of HAART was 80% lower than among non-HAART patients (adjusted RR = 0.20, 95% CI: 0.12-0.34). Cotrimoxazole prophylaxis also reduced morbidity (adjusted RR = 0.65, 95% CI: 0.45-0.94). CONCLUSION: These results confirm the reduction in morbidity due to HAART, and the additional protection of cotrimoxazole prophylaxis.