OBJECTIVES::  Adherence to highly active antiretroviral therapy (HAART) is crucial to HIV treatment effectiveness. The use of pharmacy refill data has become an increasingly important approach to measure adherence. Thus, administrative databases that contain such data could be especially useful. This is the case of SICLOM, the national system for logistic control of medicines of the Brazilian National STD/Aids Program. Despite its administrative nature, this system contains specific information about antiretroviral dispensation providing a unique data source for adherence assessment. This study aimed to assess HAART adherence among patients followed up in a public outpatient facility located in Rio de Janeiro city, Brazil, using pharmacy refill data obtained from SICLOM. METHODS:: All patients that initiated HAART in 2011 were included and followed up until December 2013. Data about dispensed medicines, including amounts and dates, and demographic data was obtained from SICLOM. Therapy adherence was estimated using the medication possession ratio (MPR) calculated as the ratio between the number of days covered by a refill and treatment duration in days. The number of days without medication (gaps) was also calculated. RESULTS::  The analysis included 82 patients. The average age was 36.2 ±10.1 years old and 85.4% were male. Most of the patients (65.9%) were followed up at public facilities. The medians of the gaps and MPR were 66.5 days (28.75-130) and 94.6% (87.9-97.4) respectively. MPR was lower than 95% for 54.9% of the patients. CONCLUSIONS::  According to the World Health Organization, HIV patients should take more than 95% of antiretroviral prescribed doses. However, the fact that 54.9% of the patients presented a MPR lower than 95% suggests a high prevalence of HAART non-adherence. SICLOM showed to be a powerful data source for adherence studies that could provide useful information in order to improve HAART effectiveness.