Objective: To prove intra- and inter-observers reliability of ultrasound (US) in the assessment of lipoatrophic findings related to the HIV associated Adipose Redistribution Syndrome (HARS). Patients and Methods: In two separated sessions, 2 consecutive measurements of subcutaneous fat thickness (SFT) were performed by each observer at the deepest point of Bichat pad, the dorsal face of arm and the mid thigh for the assessment of facial, brachial and crural lipoatrophy, respectively. We enrolled 20 HIV patients, rotating an experienced and untrained sonologist. The assessments were performed avoiding any stand off pads in the skin and excluding artefacts due to the too abundant quantity of gel to obtaining, with minimal transducer pressure, the best resolution of the reference points. Results: Means of facial, brachial and crural SFT showed no significant differences between the workers. Coefficients of variability (SD/mean x100) were similar for facial (ranges: 4.7-5.2% vs 4.9-5.6%, respectively), brachial (ranges: 5.8- 8.4% vs 9.7-11.2%) and crural SFTs (ranges: 5.9-6% vs 6.2-8.7%). There was greater consistency in the measurements performed by the experienced vs the untrained worker. Inter-observer agreement, assessed through kappa statistic (k) analysis, confirmed increased measurements agreement in the facial (k ranged from 0.40 to 0.60), brachial (k: 0.23-0.63) and crural SFT assessments (k: 0.58-0.70) from the 1st to 2nd session. Conclusions: US shows low intra observer variability and good inter observer reliability in the assessment of body fat changes related to the HARS. The different degree of consistency by the workers and the improvement of interobserver agreement, suggest to stating a well defined period of training to obtain better US reliability.