ABSTRACT.Purpose: To evaluate whether microsurgical steps in vitreoretinal surgery can be taught by a computer-assisted training system. Methods: This prospective, randomized experimental study included 14 ophthalmic residents and medical students who were completely inexperienced in microsurgery. They were randomized into two groups. The study group underwent training programmes in a computer-assisted training system for simulation of pars plana vitrectomy. The control group did not participate in any in vitro training. In the second phase of the study, participants of both groups performed a pars plana vitrectomy in three pig eyes, which included picking a metallic foreign body from the retinal surface. Results: The amount of retinal detachment and the number of retinal defects at the end of the vitrectomies were smaller, the time needed to remove the foreign body was shorter, the number of retinal lesions associated with the foreign body removal was lower, and the mark given was better in the trained study group than in the untrained group. The relatively small number of study participants did not allow the differences between the study and control groups to reach the 5% level of error probability. Conclusions: In an animal model, training by a computer-based medical work station for simulation of pars plana vitrectomy showed better outcome measures for trained study participants compared with untrained study participants. Future studies may show whether further refinements of such training programmes will result in statistically significantly better results in surgical outcome parameters.