Given that people in many jobs suffer from intense pressure being imposed on their muscles, work-related disabilities such as musculoskeletal disorders have turned into a major concern in industrial countries. Considering the significant financial and physical burden these disorders can put on people and society as a whole, preventing these issues seems more reasonable than remedying them. In this respect, there is a need for further studies concerning the prediction of muscle fatigue and activity under different working conditions. Accordingly, the present study considers an important aspect of this issue by focusing on postures in which the workers do not have access to the work station in the frontal direction. More specifically, the main purpose of this study is to present a statistical model to predict muscle fatigue, for which electromyographic signals are collected from the muscles of individuals while working at a simulated workstation, according to which the activities of the Longissimus thoracis and Iliocostalis Cervicis muscles are evaluated. Afterward, the wavelet transform is employed via Rbio 3.1 function at seven levels to process the collected signals, followed by using the normal mean absolute value index for feature extraction. Finally, some statistical models are created by the generalized estimating equation method. According to the results, posture factors, assembly cycle time, and rest intervals between cycles, which are variables, revealed significant impacts (p < .05) on muscle fatigue. It should be mentioned that the most suitable levels of the mentioned variables are also determined based on the Taguchi design of the conducted experiments. The presented statistical models can be used for designing and comparing workstations with respect to pressure on muscles for more effectively assigning workstations to employees, planning, and scheduling work cycles, and designing industrial machinery.