Little research has been published about the impact of simulation technology on the learning process of novel motor skills. Especially the role of augmented feedback (FB) on the quality of performance and the transfer of the acquired behaviour to a no-augmented FB condition require further investigation. Therefore, novice dental students were randomly assigned to one of three groups and given the task of drilling a geometrical class 1 cavity. The FB group trained under augmented visual FB conditions, provided by the virtual reality (VR) system (DentSim). The no-FB group practised under normal vision conditions, in the absence of augmented FB. A control group performed the test sessions without participating in any training programme. All preparations were evaluated by the VR grading system according to four traditional (outline shape, floor depth, floor smoothness and wall inclination), and two critical, criteria (pulp exposure and damage to adjacent teeth). Performance analyses revealed an overall trend towards significant improvement with training for the experimental groups. The FB group obtained the highest scores. It scored better for floor depth (P < 0.001), whilst the no-FB group was best for floor smoothness (P < 0.005). However, at the retention tests, the FB group demonstrated inferior performance in comparison with the no-FB group. The transfer test on a traditional unit revealed no significant differences between the training groups. Consequently, drilling experience on a VR system under the condition of frequently provided FB and lack of any tutorial input was considered to be not beneficial to learning. The present data are discussed in view of the guidance hypothesis of FB, which refers to the apprentice's dependence on FB.