Background Virtual reality is an innovative technology for medical education associated with high empirical realism. Therefore, this study compares a conventional cardiopulmonary resuscitation (CPR) training with a Virtual Reality (VR) training aiming to demonstrate: (a) non-inferiority of the VR intervention in respect of no flow time and (b) superiority in respect of subjective learning gain. Methods In this controlled randomized study first year, undergraduate students were allocated in the intervention group and the control group. Fifty-six participants were randomized to the intervention group and 104 participants to the control group. The intervention group received an individual 35-min VR Basic Life Support (BLS) course and a basic skill training. The control group took part in a “classic” BLS-course with a seminar and a basic skill training. The groups were compared in respect of no flow time in a final 3-min BLS examination (primary outcome) and their learning gain (secondary outcome) assessed with a comparative self-assessment (CSA) using a questionnaire at the beginning and the end of the course. Data analysis was performed with a general linear fixed effects model. Results The no flow time was significantly shorter in the control group (Mean values: control group 82 s vs. intervention group 93 s; p = 0.000). In the CSA participants of the intervention group had a higher learning gain in 6 out of 11 items of the questionnaire (p < 0.05). Conclusion A “classic” BLS-course with a seminar and training seems superior to VR in teaching technical skills. However, overall learning gain was higher with VR. Future BLS course-formats should consider the integration of VR technique into the classic CPR training or vice versa, to use the advantage of both teaching techniques.