In this paper, we highlight knowledge as specific channel through which labour mobility affects conditional within-firm wage dispersion. We present a model in which workers acquire knowledge on the job and firms pursue a policy of learning-by-hiring. The latter generates workers flows that connect firms in a network. A firm’s position in the network depends on its capacity to absorb the tacit knowledge developed by other firms in the economy. The model predicts that firms central to the network, those with the highest absorptive capacity of tacit knowledge, have the highest wage dispersion. Using 1995-2001 Veneto (a region of Italy) matched employer-employee data, we map workers flows between firms and build the network formed by all the firms. For each firm, we assess its network centrality. In our data conditional within-firm wage dispersion turns out to be increasing in network centrality, confirming the prediction of the model.