This paper presents a distributed algorithm capturing the TrustWebRank metric, whose goal is to evaluate trust relationships among users in a social network. We first justify why a distributed version of TrustWebRank is useful and then we assess its performances, both in terms of complexity and bandwidth, through several simulations performed using an ad-hoc tool. The scenario we used for simulations is the Epinions.com social network. Results of simulations are finally discussed highlighting pros and cons of the proposed algorithm and possible approaches to improve its performances.