Along with the issues and limitations inherent in all observational studies, occupational epidemiologic studies have to address biases arising from the healthy worker effect. This mechanism can be described as one of a time varying confounder affected by prior exposure, which cannot be addressed in standard regression approaches. Recent applications of a series of methods collectively known as ‘g-methods’, have sought to address this issue in occupational epidemiologic studies. Furthermore, some studies seek to provide direct estimates of risk (cumulative incidence) in relation to exposures of interest, under hypothetical exposure scenarios and interventions. These measures are advantageous compared to traditional measures of contrast relying on the hazard or the odds, and are of direct significance to risk and policy assessment. This talk will portray issues that may give rise to bias in occupational studies, such as time-varying confounding affected by prior exposure, right censoring, competing events, and left truncation in relation to methods used to address them. We also lay out steps in identifying target quantities of interest, given the existing knowledge and questions at hand, assessing whether that quantity is identifiable with available estimation methods, and the interpretation of these quantities after estimation.