The p – hub median problem is an NP hard location – allocation problem, that consists of finding p points to establish facilities and the assignment of the users to these points. In the capacitated version of this problem, each hub has got a maximum capacity limiting the traffic to be assigned. A new evolutionary approach that has been very effective for solving optimization problems is Path Relinking, an extension of Scatter Search that links solutions over neighborhood spaces. GRASP is a well-known randomized multistart metaheuristic. In this paper, we present a hybrid GRASP-Path Relinking for the capacitated p – hub median problem where the GRASP is used to construct the population of the Path Relinking. Computational results demonstrate that the hybrid GRASP-Path Relinking provides better solutions, in terms of both running times and solution quality.