A longitudinal test of the Job Demands‐Resources (JD‐R) model of work stress and engagement (Bakker & Demerouti, 2007; Demerouti et al., 2001) was conducted in a sample of Australian university academics (N= 296). The aim was to extend the JD‐R model by (1) determining how well job demands (work pressure, academic workload) and job resources (procedural fairness, job autonomy) would predict psychological strain and organisational commitment over a three‐year period, and (2) incorporating longitudinal tests of reversed causation. The results of SEM analyses showed that Time 1 resources directly predicted Time 2 strain and organisational commitment, but that Time 1 demands predicted Time 2 strain only indirectly via job resources. We did not find evidence for reversed causation. We discuss possible mediators of the relationships between working conditions and work stress outcomes, and the practical implications of the results.