Purpose Training within a proficiency-based virtual reality (VR) curriculum may reduce errors during real surgical procedures. This study used a scientific methodology to develop a VR training curriculum for phacoemulsification surgery (PS). Patients and methods Ten novice-(n) (performed o10 cataract operations), 10 intermediate-(i) (50-200), and 10 experienced-(e) (4500) surgeons were recruited. Construct validity was defined as the ability to differentiate between the three levels of experience, based on the simulator-derived metrics for two abstract modules (four tasks) and three procedural modules (five tasks) on a high-fidelity VR simulator. Proficiency measures were based on the performance of experienced surgeons. Results Abstract modules demonstrated a 'ceiling effect' with construct validity established between groups (n) and (i) but not between groups (i) and (e)-Forceps 1 (46, 87, and 95; Po0.001). Increasing difficulty of task showed significantly reduced performance in (n) but minimal difference for (i) and (e)-Anti-tremor 4 (0, 51, and 59; Po0.001), Forceps 4 (11, 73, and 94; Po0.001). Procedural modules were found to be construct valid between groups (n) and (i) and between groups (i) and (e)-Lenscracking (0, 22, and 51; Po0.05) and Phacoquadrants (16, 53, and 87; Po0.05). This was also the case with Capsulorhexis (0, 19, and 63; Po0.05) with the performance decreasing in the (n) and (i) group but improving in the (e) group (0, 55, and 73; Po0.05) and (0, 48, and 76; Po0.05) as task difficulty increased. Conclusion Experienced/intermediate benchmark skill levels are defined allowing the development of a proficiency-based VR training curriculum for PS for novices using a structured scientific methodology.