The paper develops a dynamic antenna scheduling strategy for downlink MIMO communications, where the transmitted signal for each user is beamformed towards a selected subset of receive antennas at this user. The proposed method removes the condition on the number of transmit-receive antennas in comparison to traditional zero-forcing and time-scheduling strategies. By characterizing the probability distribution of the so-called signal-to-leakage-plus-noise (SLNR) ratio, we show that there is an optimal set of receive antennas that maximizes the system performance for each channel realization. This fact is used to propose an antenna scheduling scheme that leads to improvements in terms of SINR outage probabilities.