In this paper we discuss the contextual support provided in TM4L - an environment for building and using Topic Maps-based learning repositories. We propose an extension of the Topic Maps model with contexts and an approach for contextual retrieval of information. Our approach is based on defining context primitives and a simple language for expressing context queries. The resources retrieved by running the queries can be directly examined or used as seeds for further locating relevant resources in the repository. The contextual retrieval functionality is provided through an intuitive user interface. We discuss the context representation and the contextual retrieval implementation in TM4L.