Purpose: To investigate the learning curve of robot-assisted vitreoretinal surgery compared to manual surgery in a simulated setting. Methods: The study was designed as a randomized controlled longitudinal study. Eight ophthalmic trainees in the 1st or 2nd year of their specialization were included. The participants were randomized to either manual or robot-assisted surgery. Participants completed repetitions of a test consisting of three vitreoretinal modules on the Eyesi virtual reality simulator. The primary outcome measure was time to learning curve plateau (minutes) for total test score. The secondary outcome measures were instrument movement (mm), tissue treatment (mm²) and time with instruments inserted (seconds). Results: There was no significant difference in time to learning curve plateau for robot-assisted vitreoretinal surgery compared to manual. Robot-assisted vitreoretinal surgery was associated with less instrument movements (i.e. improved precision), −0.91 standard deviation (SD) units (p < 0.001). Furthermore, robot-assisted vitreoretinal surgery was associated with less tissue damage when compared to manual surgery, −0.94 SD units (p = 0.002). Lastly, robot-assisted vitreoretinal surgery was slower than manual surgery, 0.93 SD units (p < 0.001). Conclusions: There was no significant difference between the lengths of the learning curves for robot-assisted vitreoretinal surgery compared to manual surgery. Robot-assisted vitreoretinal surgery was more precise, associated with less tissue damage, and slower.