Abstract-This paper aims to investigate the extent to which the use of a virtual patient (VP) improves the level of theoretical knowledge and confidence (self-efficacy) of psychology and medicine students to perform clinical diagnostics. We created two classes on an e-learning Moodle platform, presenting as a virtual patient the evaluation of the same panic disorder case. One class used a linear VP, the other, a branched VP (a decision-tree type). The sample consisted of 159 volunteer participants, a mix of psychology and medicine students, randomly assigned to one of three conditions: control, linear virtual patient (PVL) and branched virtual patient (PAD). Before and after the training with the virtual patient sequence, a questionnaire investigated the theoretical knowledge and the self-efficacy levels. For self-efficacy evaluation, we used the Session Management sub-scale of Counsellor Activity Self-Efficacy Scales, by Robert W. and Clara E. Hill.Both types of VPs are associated with a statistically significant increase of knowledge about panic disorder diagnostic criteria, reinforcing the fact that the virtual patient has an impact on the cognitive profile (declarative knowledge) and on self-efficacy (related to procedural knowledge). Still, the study does not prove that one of the two types of virtual patient have a significant better outcome.