A buffer sharing scheme need to meet many competing objectives like robustness to changes in traffic conditions, fairness, efficiency, implementation simplicity, etc. Of the existing techniques, Push Out (PO) is optimum in terms of overall cell loss performance, but its delay performance is worse than that of all other methods for the multiple space priority. Another scheme named Dynamic Threshold (DT) is robust and adaptive which has good delay performance. A hybrid technique named Dynamic Threshold with native Push Out (DnP) shows the best delay performance but its loss performance is worse specially for higher priority traffic. A novel policy called Balanced Buffer Push Out (BBPO) is proposed here that aims to achieve a better loss performance than DT and better delay performance than PO for multipriority traffic. Using computer simulation the loss performance and delay performance of DT, PO, DnP and BBPO are measured for bursty traffic, and a comparison of the results is presented. The proposed sche A buffer sharing scheme need to meet many competing objectives like robustness to changes in traffic conditions, fairness, efficiency, implementation simplicity, etc. Of the existing techniques, Push Out (PO) is optimum in terms of overall cell loss performance, but its delay Another scheme named Dynamic Threshold (DT) is robust and adaptive which has good delay performance. A hybrid technique named Dynamic Threshold with native Push Out (DnP) shows the best delay A novel policy called Balanced Buffer Push Out (BBPO) is proposed here that aims to achieve a better loss performance than DT and better delay performance than PO for multipriority traffic. Using computer simulation the loss performance and delay performance of DT, PO, DnP and BBPO are measured for bursty traffic, and a comparison of the results is presented. The proposed sche