The objective of the research was to analyze the internal structure of the Rosenberg Self-Esteem Scale (RSE) and to verify the differential functioning of the items according to sex. An instrumental study was developed in two samples, the first one consisting of 450 (226 women and 224 males) and the second one by 481 (225 females and 256 males) Peruvian adolescents belonging to public and private educational institutions of Metropolitan Lima. The internal structure was evaluated by exploratory and confirmatory factor analysis. Internal consistency analysis was performed, which for all models was good (H>.80). On the other hand, differential functioning was estimated through factor invariance. The results indicate that the RAS should be interpreted according to a two-dimensional model, with a strong partial invariance, and items 1, 5 and 10 should be released, suggesting that item 8 be removed. Based on the results of the study, RSE has empirical evidence of validity, reliability and non-invariant operation of some items.