This paper presents a performance-based approach to the study of the phonetic encoding of sexual identity in Italian. Six male speakers were recruited after being controlled for familiarity, interaction, and comfort with members of both hetero- and homosexual communities. They were recorded reading introductory, scientific, and narrative paragraphs with the goals of conveying straight and gay identities, respectively, as well as in an unmarked style. Analysis focused on segmental (vowels, /s/, liquids), suprasegmental, and pragmatic components of performances, the results of which suggest that speakers use a combination of phonetic features to instantiate socially salient identities.