The study evaluated the use of sonographic measurements as an alternative to assessments based on clinical or other imaging techniques for the diagnosis of body-fat abnormalities. The study enrolled 179 HIV-infected patients, 81 (45.3%) of them diagnosed as lipodystrophy (LD)-positive based on a clinical standard questionnaire. Association between clinical LD and sonographic measurements of face, right upper limb, subcutaneous abdomen, and visceral compartments was evaluated by multiple logistic regression. The predicted probability of the logistic model was 0.64, corresponding to a maximum sensitivity of 69.1% (58%-79%), a specificity of 94.9% (88%-98%), and to positive and negative predictive values of 92% (82%-97%) and 79% (70%-86%), respectively. Kappa measure of concordance was 65% (54%-77%). Low sensitivity poses a problem for the use of sonography to detect LD in the clinical routine as a single exam, speaking in favor of the combined use of clinical and sonographic measurements over time.