We used a full-panel longitudinal design to investigate, separately for women and men, the hypotheses that changes in the components of the Job Demands-Control-Support (JDC-S) model predict changes in depression symptoms levels over time and that the reversed prediction would also be found. Our study was conducted on a multi-occupational sample of apparently healthy employees (N = 692, 68% men) using three waves of data gathering, replicating our tests on two time lags of 18 months and 3 years on average. We controlled for neuroticism and other potential confounding variables. For both time lags, support for our hypotheses was found for the men only. We did not find systematic differences between the time lags, nor did we find a predominance of one of the unidirectional effects examined. We outline the theoretical and practical implications of our findings, including their relevance for efforts to combat depressive symptoms by changing job characteristics.