The relationship between characteristics of HIV antiretroviral regimens and treatment adherence was studied in adolescent and adult patients who underwent antiretroviral therapy from January 1998 to September 2000, at the Service for Specialized Assistance in Pelotas. The patients were interviewed on two occasions, and the use of antiretrovirals during the previous 48 hours was investigated by a self-report. Adherence was defined as use of 95% or more of the prescribed medication. Social-demographic variables were collected through direct questionnaires. The antiretroviral regimen and clinical data were copied from the patients' records. Associations between the independent variables and adherence were analyzed by means of logistic regression. The multivariate analysis included characteristics of the antiretroviral regimens, socialdemographic variables, as well as perception of negative effects, negative physiological states, and adverse effects of the treatment. Among the 224 selected patients, 194 participated in our study. Their ages varied from 17 to 67 years; most patients were men, with few years of schooling and a low family income. Only 49% adhered to the treatment. Adherence to treatment regimens was reduced when more daily doses were indicated: three to four doses (odds ratio of adherence to treatment (OR)=0.47, 95% confidence interval (CI) 0.22-1.01) and five to six (OR=0.24, 95% CI 0.09-0.62); two or more doses taken in a fasting state (OR=0.59, 95% CI 0.11-0.68), and for patients who reported adverse effects to the treatment (OR=0.39, 95% CI 0.19-0.77). Most of the regimens with more than two daily doses of medication included at least one dose apart from mealtimes. The results suggest that, if possible, regimens with a reduced number of doses should be chosen, with no compulsory fasting, and with few adverse effects. Strategies to minimize these effects should be discussed with the patients.