This paper compares two broad views of psychotherapy, the etiological and the rehabilitative. The etiological view is based on the premise that the psychotherapeutic process provides a basis for laying bare the causal link between 'deeply repressed' unconscious forces and the patient's current condition. Spence's critique of this position is based on drawing a distinction between narrative truth and historical truth, pointing out the persuasive or rhetorical power of a coherent synthesis of the material presented by the patient, regardless of its historical veracity. Michels also draws attention to this difficulty by pointing to Freud's discovery that reports from childhood were based not on fact, but on fantasies which re-emerge in the transference relationship. The proposal that the therapist provides a context more conducive to the resolution of childhood trauma implies that the trauma was fact and not fantasy. Frank has drawn the parallel between confession in a cueless vacuum with brainwashing techniques which result in the confession of non-events. Besides these difficulties, the etiological approach cannot establish the direction of supposed causality linking childhood events, repressed psychological forces and the patient's condition. The apparently consistent pattern of expectations, perceptions, behavior and 'defenses' may themselves be the effects rather than the causes of the patient's condition. The psychotherapeutic process is intrinsically incapable of deciding objectively between these possibilities. Grunbaum has emphasized this epistemological weakness in the etiological position. The author therefore proposes a more pragmatic, rehabilitative, view of the psychotherapeutic process. Psychotherapy that deals with the personal and social life of the patient may help to alter the circumstances that contribute to discomfort, or change behavior that leads to dissatisfaction, without any commitment to unproven etiological theories. In this sense, psychotherapy, like physiotherapy and rehabilitation, is non-specific and can be applied without making any assumptions regarding specific causal conditions or pathogenetic factors.