The discrete Fourier transform (DFT)-based codebook is employed in this paper to quantize channel state information so that the amount of feedback can be reduced in the multiple input multiple output (MIMO) downlink of long term evolution (LTE) system.And a novel beamforming (BF) scheme based on the proposed channel quality-to-interference (QIR) quantizing criteria is developed,which uses only the index of the optimal codebook for the beamforming at the base station (BS),and dramatically reduces the amount of feedback.The proposed BF scheme jointly considers the influences of the quality of the quantized channels and the mutual interference among the sub-channels.The extensive simulation results verify that throughput of the proposed BF scheme is better than that of the random BF with a little feedback,and that of the eigen-beamforming even under low signal noise ratio (SNR) scenario.