The Brazilian public health system has implemented free, universal access to antiretroviral (ARV) therapy for HIV-infected patients. To evaluate this system, we performed a pilot study to determine whether ARVs were prescribed according to Brazilian guidelines in place in 2000, and whether prescriptions were refilled in a timely manner. Year 2000 data were abstracted from all medical and pharmacy records of adult patients first registered for HIV/AIDS care in a Rio de Janeiro public clinic from January to June 2000 (n = 67). Results were analysed using frequency analyses, chi-square tests and logistic regression. The patient sample was 41.8% female and had a mean age of 34.9 years. 54 (81%) had AIDS; total sample mean baseline CD4+/viral counts were 276 cells/mm3 and 237 517 copies per millilitre, respectively. Delays between clinic request and receipt of first CD4+/viral load results ranged from 25 to 107 (mean 66) and 33 to 139 (mean 86) days, respectively. Fifty-nine patients (88.1%) were prescribed ARV treatment. Forty-two regimens (71.2%) were highly active antiretroviral therapies; 17 (28.8%) were combination regimens with two nucleoside reverse transcriptase inhibitors. No combinations were prescribed that were contraindicated in Brazilian guidelines, however 33 patients (55.9%) were prescribed ARV drugs before one or both HIV status parameters (initial CD4+ level or viral load) were recorded. Fourteen patients prescribed ARVs (23.7%) lacked a supply of medication for >1 month during the year at least once. Of these patients, 11 had treatment lapses as a result of failure to pick up medications, and three lacked medication because of drug shortages. Medication lapses were associated with female sex, being hospitalized in 2000, and having more than two drugs in regimen, but were not associated with age, CD4+ level or use of ARVs before 2000. The results from this pilot study suggest conservative prescription of HAART, high practitioner adherence to guidelines, and some problems with refilling medications in a timely manner. Monitoring delays were identified as a structural limitation to optimal adherence to practice guidelines. Better access to monitoring-laboratory facilities and greater drug availability would improve programme success.