The nearby-friend problem for a user in the context of location based services is to learn the location of a nearby-friend only if (s)he is actually nearby, while simultaneously maintaining the privacy of their respective locations. The existing works address the problem for two users only. In this paper, we present for the first time a solution which caters to multiple users. We use techniques from Euclidean plane geometry and graph theory to develop a secure multi-party computation approach towards solving the nearby-friend problem efficiently for many users. Our solution ensures users' location privacy by enforcing k-anonymity.