The invention involves local media rendering of a multi-party call, performed by a Client User Equipment (1). The media is encoded by each party in the call, and sent as a media stream to a Media server (2), and the media server receives a request for media streams from each Client User Equipment, each media stream in the request associated with a client priority. The Media server selects the media streams to send to each Client User Equipment, based on the request, and further such that the number of streams does not exceed a determined maximum number, which is based e.g. on the available bandwidth.