We develop a queuing analytic model to study cross-layer effects on Quality of Service (QoS) performance for downlink transmission in a Vertical Bell Laboratories Layered Space-Time Architecture (V-BLAST) Multiple Input Multiple Output (MIMO) wireless system exploiting multiuser diversity. As in multiuser Single Input Single Output (SISO) systems, multiuser diversity has been shown to have a great potential in improving system throughput in multiuser MIMO systems as well. In a V-BLAST MIMO system, the transmit antennas can carry parallel streams and multiple users can be scheduled for transmissions at the same time. We consider a multiuser diversity scheme that can effectively exploit this extra dimension in multiuser scheduling. To investigate the cross-layer aspect of the performance improvement, we perform a queuing analysis to derive buffer statistics, queuing delay distribution, packet throughput and loss rate experienced in the data link layer when this multiuser diversity technique is deployed in the system. We also present selected numerical results to show how the queuing model can help us to relate these important QoS measures to relevant physical layer and traffic parameters. Usefulness of the developed analytical model is also demonstrated through example applications.