In this paper we show that Gaussian sum particle filters (GSPFs) present an effective solution for the problem of estimating phase noise in digital communications. We start by describing the problem of estimating the phase noise in terms of its general Bayesian formulation and then present the solution offered by the GSPF. Our filter is based in two central features. On the one hand, the system model comprising both the dynamics model and the observations model and on the other hand the sensor factor, a periodic function with respect to the current observation of the state. Although phase noise estimation algorithms are widely rep- resented in the literature to our knowledge it has never been considered to use GSPFs for this purpose. Furthermore, other solutions using conventional particle filters assumed the modu- lation symbols to be PSK, i.e., phase-defined, while we assume phase- and amplitude-defined symbols, e.g., M-QAM.