Intervocalic s -voicing in Italian has been frequently used to motivate or illustrate theoretical concepts, such as the prosodic word or base-output correspondence. In this paper, I provide a detailed examination of the pattern in two varieties of Italian, Lombardian and Tuscan. This closer look at the data reveals problems for previous accounts. I will provide an analysis of the pattern within Optimality Theory and show that derived environment effects can be accounted for by recourse to the featural level in the interpretation of alignment constraints and by the use of a lesser studied faithfulness constraint, i.e., the contiguity constraint. Furthermore, I will give an account of the fact that only s shows this pattern while all other consonants are excluded.