OBJECTIVES: Although figures on the evolution of prescriptions of ARV are now available, little is known about the dynamics and reasons of treatment changes. The STAR cohort was initiated in February 2001 across 3 French HIV Public Hospital to monitor and to explain the evolution of ARV therapies. METHODS: STAR is an observational study implemented to prospectively collect electronic data on anti-HIV treatments, immunological and virological status and medical cost of a random sample of patients. It aims at identifying therapeutic strategies in the setting of clinical practice. RESULTS: A total of 1177 patients (77% male) were included, of mean age 41±9 years. A total of 290 patients (24.6%) had clinical AIDS. Mean CD4 counts and viral load were respectively 478±275 /mm3 and 29,300±89,400 copies/ml in February 2001 versus 506±283 /mm3 and 15,300±55,600 copies/ml in December 2002. Of the whole cohort, 17% were ARV naïve in February 2001, among whom 52.5% initiated an ARV treatment during the observation period. The most frequent association in February 2001 was 2 Nucleoside Reverse Transcriptase Inhibitors (NRTI) + 1 Protease Inhibitor (PI) with 24.2% of subjects, and in December 2002 2 NRTIs + 1 Non Nucleoside Reverse Transcriptase Inhibitor (NNRTI) with 21.4%. Among emerging treatment schemes, boosted PIs raised from 13.8% to 24.1%. Sixty percent of patients had their regimen changed at least once over the 21-month period. Main reasons of treatment changes were toxicity (22%) and failure (21%). Among treatment discontinuations, 32% were attributable to supervised treatment interruption, 24% to toxicity and 14% to patient's decision. CONCLUSIONS: These preliminary results obtained after 21 months of follow-up of the STAR cohort show significant trends in the changes of treatment regimens with more simplified regimens and more frequent multiple therapies. These findings could help policy makers in the elaboration of guidelines and the evaluation of adherence behavior.