Methods A retrospective cohort of 124 patients from the Parirenyatwa Hospital Kaposi’s Sarcoma and Opportunistic Infections (OI) Clinics was studied. 31 patients with AIDS-KS were matched 1:3 to 93 non-KS AIDS patients based on date of initiation of HAART, gender, and age. The primary endpoint was loss to care, defined as failure to attend clinic or refill prescriptions for 3 months or longer. Secondary endpoints were weight gain at 6 months, change in CD4+ count within 1 year, and final CD4+ count within 1 year of initiating HAART. Eligibility criteria included a minimum of 6 months of followup in the OI Clinic and less than 2 months of previous HAART prior to beginning therapy through the OI Clinic. A two-step model-selection strategy using KS status, gender, age, WHO performance status, OI disease burden, medical aid, employment, education, pretreatment cotrimoxazole use, pre-treatment weight, and pre-treatment CD4+ count was used to identify factors associated with loss to care. On the initial univariate analysis, KS status, medical aid, and pre-treatment cotrimoxazole use had a p<0.15 and were included in the final multivariate analysis. Results AIDS-KS and non-KS patients did not differ significantly in baseline characteristics except for pre-treatment CD4+ count (196 vs. 92 cells/mm, p=0.005). On the multivariate analysis, KS status (p=0.016, HR: 4.11, CI: 1.31-12.92) and having medical aid were significant predictors of loss to care (p=0.048, HR: 3.84, CI: 1.0214.44). At a median followup of 632 days, 37.5% of AIDS-KS patients were lost to care compared to 16.1% of non-KS patients. AIDS-KS patients had significantly worse weight gain than non-KS patients (+0.78% vs. +4.18%, p=0.023). Change in CD4+ count (p=0.149) and final CD4+ count (p=0.729) were not significantly different between study groups. Amongst AIDS-KS patients, retained patients (n=20) had significantly higher pre-treatment CD4+ counts than patients lost to care (n=11) (232 vs. 122 cells/mm, p=0.048).