Work-related musculoskeletal disorders are often associated hazardous body postures and repetitive tasks. To ensure good working conditions, workplaces must be ergonomically designed. Risk assessments are used to identify inadequate workplace design. Sometimes this can a time-consuming process when analyzing the movements of workers and handled loads. In particular, it is currently difficult to assess the external forces acting upon workers. The use of wearable sensor technology for ergonomic assessment could assist ergonomists in this task in the near future. The emerging field of machine learning opens up new possibilities for data processing. A case study was conducted to investigate the extent to which external forces, which normally require complex measurement technology, can be estimated from kinematic data alone. For this purpose, a full-body kinematic (17 portable IMU sensors) was used. A subject performed 1125 lifts with 5 different heavy boxes (0–20 kg). The support vector machine classifier showed an average accuracy of over 90% for the test data set. The speed of the forearms towards the body, upper body movement and the flexion of the wrist provided the most important information for the classification models. The results indicate that weight classification under standardized conditions can be performed only on the basis of kinematic data and an individual model. However, large data sets are needed to generalize the results.