This study investigated the moderating effects of social support by supervisors and colleagues relative to social stressors at work and depressive symptoms using a structural equations approach in a 3-wave longitudinal study over 1 year. The analyses were based on a randomly drawn sample (N = 543) of citizens in the area around Dresden in the former East Germany. LISREL analysis with latent moderating effects revealed a moderating effect for supervisor support. This applied only if the time lag was 8 months, but not for longer or shorter lags. Under low-support conditions depressive symptoms were increased by social stressors, whereas, contrary to expectations, social stressors reduced subsequent depressive symptoms under high-support conditions. No moderating effect for colleague support was found. Several mechanisms are discussed that may explain the results.