Despite the increased use in sports, it is still unclear to what extent VR training tools can be applied for motor learning of complex movements. Previous VR studies primarily relate to realize performances rather than learning motor skills. Therefore, the current study compared VR with video training realizing the acquisition of karate technique, the Soto Uke moving forward in Zenkutsu Dachi, without being accompanied by a trainer or partner. Further analyses showed whether a less lavished forearm compared to a whole-body visualization in VR is necessary to acquire movements’ basics sufficiently. Four groups were tested: 2 groups conducted VR training (VR-WB: whole-body visualization, and VR-FA having only visualized the forearms), the third group passed through a video-based learning method (VB), and the control group (C) had no intervention. In consultation with karate experts, a scoring system was developed to determine the movements’ quality divided, into upper- and lower body performance and the fist pose. The three-way ANOVA with repeated measurements, including the between-subject factor group [VR-WB, VR-FA, VB, C] and the within-subject factors time [pre, post, retention] and body regions [upper body, lower body, fist pose], shows that all groups improved significantly (except for C) with the similar course after four training sessions in all body regions. Accordingly, VR training seems to be as effective as video training, and the transfer from VR-adapted skills into the natural environment was equally sufficient, although presenting different body visualization types. Further suggestions are made related to the features of future VR training simulations.