In this paper, we investigate the beamforming optimization for the downlink of multiuser MISO (MU-MISO) system with finite-rate feedback, where the transmitter is only aware of a delayed and quantized version of the channel state information (CSI). By taking into consideration both CSI quantization error and feedback delay, we propose an improved beamforming optimization criterion by maximizing the signal-to-leakage-and-noise ratio (SLNR) averaged over CSI imperfection. A closedform expression for the optimized beamforming design is then presented. Numerical results verify that the proposed scheme is more robust to the CSI imperfection in finite-rate feedback system than conventional SLNR-based beamforming scheme as well as another well-known zero-forcing beamforming (ZFBF) scheme in terms of both sum rate and bit error rate (BER).