This study aimed at understanding the ability of a VR-based assembly tasks simulator to evaluate physical risk factors in ergonomics. Sixteen subjects performed simplified assembly tasks in RE and VE. Motion of the upper body and 5 muscle electromyographic activities were recorded to compute normalized and averaged objective indicators of discomfort, i.e Rapid Upper Limb Assessment score, Averaged Muscle Activations and Total Task Time. Rated Perceived Exertion (RPE) and a questionnaire were used as subjective indicators of discomfort. Timing regime and complexity of the assembly tasks were investigated as within-subject factors. Results revealed significant differences between measured indicators in RE and VE. While objective measures indicated a lowered activity and exposure under VE, the subjects experienced more discomfort than in RE. Fairly good correlation levels were found between RE and VE for six of the objective indicators. This study clearly demonstrates that ergonomic studies of assembly tasks using VR are still challenging. Indeed, objective and subjective measurements of discomfort that are usually used in ergonomics to minimize the risks of Work-Related Musculoskeletal development exhibit opposite trends in RE and VE. Nevertheless, the high level of correlation found during this study indicates that the VR-based simulator can be used for such assessment.