OBJECTIVES: The aim of this study was to develop a strategy to identify adverse drug events associated with drug-drug interactions by analyzing the prescriptions of critically ill patients. METHODS: This retrospective study included HIV/AIDS patients who were admitted to an intensive care unit between November 2006 and September 2008. Data were collected in two stages. In the first stage, three prescriptions administered throughout the entire duration of these patients’ hospitalization were reviewed, with the Micromedex database used to search for potential drug-drug interactions. In the second stage, a search for adverse drug events in all available medical, nursing and laboratory records was performed. The probability that a drug-drug interaction caused each adverse drug events was assessed using the Naranjo algorithm. RESULTS: A total of 186 drug prescriptions of 62 HIV/AIDS patients were analyzed. There were 331 potential drug-drug interactions, and 9% of these potential interactions resulted in adverse drug events in 16 patients; these adverse drug events included treatment failure (16.7%) and adverse reactions (83.3%). Most of the adverse drug reactions were classified as possible based on the Naranjo algorithm. CONCLUSIONS: The approach used in this study allowed for the detection of adverse drug events related to 9% of the potential drug-drug interactions that were identified; these adverse drug events affected 26% of the study population. With the monitoring of adverse drug events based on prescriptions, a combination of the evaluation of potential drug-drug interactions by clinical pharmacy services and the monitoring of critically ill patients is an effective strategy that can be used as a complementary tool for safety assessments and the prevention of adverse drug events.