The association between the psychosocial work environment, including job demands, job control, and organizational justice, and employee wellbeing has been well established. However, the exposure to adverse work environments is typically measured only using self-reported measures that are vulnerable to reporting bias, and thus any associations found may be explained by reverse causality. Using linear regression models and cross-lagged structural equation modeling (SEM), we tested the direction of the association between established job stress models (job demand control and organizational justice models) and 3 wellbeing indicators (psychological distress, sleeping problems, and job satisfaction) among 1524 physicians in a 4-year follow-up. Results from the longitudinal cross-lagged analyses showed that the direction of the association was from low justice to decreasing wellbeing rather than the reverse. Although the pattern was similar in job demands and job control, a reciprocal association was found between job control and psychological distress.