This paper analyses the users’ group interests by mining the internet browsing history. To count the visiting information of the interests’ categories, visiting time and the number of users, get the regularity of conclusion. Then, it has put forward an improved HAC (hierarchical agglomerative clustering) and k-means algorithm to cluster the users by their interests, to mine the users’ access mode. Finally, it has proved the stability of users’ dominant interests. That means the users’ most important interests are stable as the time increases.