Issues associated with and specific aspects of cohort studies of occupational related low back pain (LBP) were described. A cohort study may be retrospective, prospective or a hybrid of the aforementioned. Cohort membership can be fixed, where the membership is static due to their presence at the same event, or dynamic, where the population gains and loses members during the study period. Sample size estimation, issues involved in definition, measurement, and models in low back events were reviewed. Acute injury and cumulative injury or pain models were compared, with neither category of model being satisfactory alone, but instead, using a unified model that combines both the acute and chronic paradigms. Nonparticipation or selection bias, information bias, the healthy worker effect and confounding effects in cohort studies were reviewed. Methods for determining and measuring back injury risk factors and exposures were discussed, and included ergonomic assessments of employee exposure in each task and job title, historical data review of job titles and activities, and investigator inspection and discussion with company management. The comparability and validity of outcome measures, the followup and/or monitoring of outcome, measures of effect and data analysis, and the interaction of exposure and worker risk factors were also discussed.