This paper proposes a method that can discover users' frequent access patterns underlying users' browsing Web behaviors. Firstly, the paper introduces the concept of access pattern according to a user's access path, and secondly puts forward a revised algorithm (FAP-Mining) based on the FP-tree algorithm to mine frequent access patterns. The new algorithm first constructs a frequent access pattern tree and then mines users' frequent access patterns on the tree. The algorithm is accurate and scalable for mining frequent access patterns with different lengths.