This paper addresses collaborative learning in the medical domain. In particular, it focuses on the evaluation of a component specially devised to promote collaborative learning using AMPLIA. AMPLIA is an intelligent multi-agent environment to support diagnostic reasoning and the modeling of diagnostic hypotheses in domains with complex, and uncertain knowledge, such as the medical domain. Recently, AMPLIA has been extended with a new component providing support in workgroup formation. Workgroups are proposed based on individual aspects of the students, such as learning style, performance, affective state, personality traits, and also on group aspects, such as acceptance and social skills. The paper also presents and discusses the results of an experiment evaluating the performance of workgroups composed according to suggestions provided by the system.