This article investigates the relationship between social stressors, comprising conflicts with co-workers and supervisors and social animosities at work, irritation and depressive symptoms. It is argued that only a few mediation hypotheses have been investigated in organizational stress research. In the present study it was hypothesized that irritation mediates the effect of social stressors on depressive symptoms. This hypothesis was tested using four waves of a six-wave longitudinal study based on a representative sample (N =313) of the residents of Dresden, Germany. The advantages of longitudinal designs were comprehensively used including the testing of different time lags, the testing of reversed causation, and modelling of unmeasured third variables that may have spuriously created the pattern of observed relationships. Structural equation modelling provided evidence for the proposed mediation mechanism and suggests that time lags of at least 2 years are required to demonstrate the effects.