There is a multiplicity of etiological factors in human cancer; psychic factors are probably neither the most important nor the most constant. Nonetheless, owing to the potential infinity of psychological variables, it is necessary to formulate hypotheses in an operational manner. The casuistic method allows a thorough clinical study of objective events, of the manner in which the patient lives them and, especially, of what they mean to him. Such an approach, however, has shortcomings which must be accepted: it cannot administer proof of the hypotheses it advances. The superiority of prospective methods over retrospective methods seems indisputable for there is no evidence that personality traits are not modified by the fatal threat of a serious illness. This criticism also holds for the modified retrospective methods, the so-called predictive methods. Research focussed on particular cancers with a study of limited psychological variables seems desirable.