In this paper, we describe a Web log mining approach to recommend learning items for each active learner based on the learner’s historical learning track. The proposed method is composed of three parts: discovering content-related item sets by Collaborative Filtering (CF), applying the item sets to Sequential Pattern Mining (SPM) and generating sequential pattern recommendations to learners. Different from other recommendation strategies which use CF or SPM separately, this paper combines the two algorithms together and makes some optimizations to adapt them for E-learning environments. Experiments are conducted for the evaluation of the proposed approach and the results show good performance of it.