This paper provides a critical review of the current explanatory models of the post-traumatic stress disorder (PTSD). Most of these models are incomplete and lack empirical confirmation. It is argued that unidimensional treatment modalities, derived from such models, fail to address the complex nature of the disorder and are likely to generate partial results. A broad hypothesis-testing approach, combining biological, psychological and psychosocial interventions, should be preferred in clinical practice. Rehabilitation should replace curative techniques and unrealistic goals in many cases.