This paper presents a Q-learning based scheme for managing the partial coverage problem and the ill-effects of free riding in unstructured P2P networks. Based on various parameter values collected during query routing, reward for the actions are computed and these rewards are used for updating the corresponding Q-values of peers. Thus, the routing is done through only nodes which have shown high performance in the past. Simulation experiments are conducted in several times and the results are plotted. Results show that the proposed scheme effectively manages free riders, generates high hit ratio, reduces network traffic and manages partial coverage problem.