The essence of Web prefetching is the construct of prediction algorithm and prefetching control. And the heart of Web prefetching is the prediction model, which is mainly classified into two main categories according to the data structure taken into account to make the prediction, namely PPM model and DG model. Although PPM model has higher predictive accuracy, it occupies large storage space. Although DG prediction algorithm can overcome PPM’s shortcoming, it can’t distinguish the initial weight of different order and do not take into account how web pages are structured, so its prediction accuracy is lower. Therefore, an improved prediction model is proposed to remedy the above defects. Experimental results show that the improved model can save the storage space and enhance higher prediction precision for the pages which have a high amount of embedded objects