BACKGROUND AND STUDY AIMS: Experienced endoscopic surgeons have adapted to the absence of depth perception while using two-dimensional (2-D) visualization. However, three-dimensional (3-D) vision may prove useful at least at the beginning of the learning curve in celioscopic training. METHODS: In a pelvitrainer with a fixed camera, two skill tests were designed to assess the performance of three groups of operators: "non-surgeons", "non-celioscopist surgeons", and "trained celioscopists". In the first test, the candidate had to touch with a needle a sequence of dots distributed on a 7.1-cm2 area. In the second test, a 6-0 C-1 needle had to be passed consecutively through two 1-mm holes made in a thin vertical plastic wall. Each test was performed ten times, using either 2-D vision (five times) or 3-D vision (five times) interspersed in a random manner. RESULTS: In both tests and within each group, performance was related to the experience of the operator, with the trained celioscopists' group obtaining the best results and the non-surgeons the worst. In every situation, including the trained celioscopists' group, 3-D vision significantly improved performances. No significant difference was observed between the results of the non-celioscopist surgeons' group using 3-D vision and those of the trained celioscopists' group using 2-D vision. CONCLUSIONS: 3-D vision improves the performance and accuracy of endoscopic surgeons. It provides a visual perception "close to reality", and helps celioscopic beginners to accelerate their training.