Conventional zero-forcing precoding was well investigated, but it hadn't considered the channel quality indicator (CQI) mismatch for practical system with limited feedback. This paper studies a novel zero-forcing precoding with channel quality indicator (CQI) estimation and adjusting, channel directional information(CDI) quantization, user scheduling and link adaptation and then proposes a CQI estimation algorithm for multiuser multiple input multiple output antenna (MU-MIMO) downlink systems with limited feedback. Simulation with 3GPP long term evolution system demonstrates that the scheme with proposed method can significantly improve system performance, compared with conventional zero-forcing precoding and single user MIMO systems.