The present study aimed at exploring the dimensionality of the Italian version of the UCLA Loneli-ness Scale-version 3 (UCLA LS3), in relation to self-esteem (Rosenberg Self-Esteem Scale — RSES), social anxiety (Social Interaction Anxiety Scale — SIAS), and adult attachment (Attachment Style Questionnaire — ASQ), in 350 Italian young adults. An innovative Exploratory Structure Equation Modeling approach (ESEM — Asparouhov and Muthén, 2009) was used. Thanks to the combination of explorative (EFA) and confirmative (CFA) factor analysis methods, ESEM allowed to simultaneously estimate an EFA measurement model with rotations and a traditional SEM model, to investigate UCLA LS3 latent structure and convergent validity. A three-factor ESEM model presented a satisfactory fit to the data. The Italian UCLA LS3 scale resulted to be composed by the interrelated dimensions of Isolation, Relational Connectedness and “Trait” Loneliness. ESEM structural part showed ASQ subscales and SIAS to systematically predict loneliness dimensions, whereas the RSES affected only the Isolation factor.