Nowadays, a wide range of initiatives are focused on providing solutions relating to the integration and reuse of different types of learning objects. This paper deals with the assembly of learning objects, one of the most difficult components for reuse. Our proposal takes into account the requirements and competencies defined for each learning object in order to allow them to be assembled in a coherent way. As a result of this approach, we have developed "ELO-Tool," a web application that implements an assembly mechanism based on a semantic comparison of requirements and competencies through the use of ontologies. Moreover, the application generates the appropriate set of meta-data to describe the new learning objects resulting from the assembly process. (Contains 3 tables and 7 figures.)