In a cross-sectional study, the relationship between physical load and musculoskeletal complaints was examined among operators (n = 77), office workers (n = 52), and miscellaneous workers (n = 15) in a tank terminal company. Information about history of musculoskeletal complaints, individual characteristics, and working conditions in past and present was obtained by a standardized interview. Assessment of physical load was performed by direct observation of awkward postures and forceful exertions during normal activities. Workers filled out a diary regarding the executed tasks during one shift. Physical load was also modeled by linking the average physical load of workers performing a particular task with the task distribution of each worker. The modeling approach was beneficial to estimate infrequent characteristics of physical load, whereas the observations were better for assessing frequent exposure to awkward trunk postures. After adjustment for age and other confounders the observed work time with trunk flexion and rotation was associated with the occurrence of back pain. The modeled work time with arms raised above shoulder showed a relationship with elbow pain, and the modeled work time with forceful exertions over 100 N showed relationships with elbow pain and wrist pain. The described relationships demonstrate that the assessment strategy chosen in a particular study strongly influences random measurement error of physical load, and that inappropriate strategies may mask true associations between physical load and musculoskeletal complaints. Modeling strategies offer the opportunity to assess physical load in dynamic work environments.