Abstract Comparison of the N.S.Q. and S.A.F. scores of women undergoing radio-therapy for a variety of malignant conditions with those of a control group reveals that the malignant group show significantly more ego-weakness (S.A.F.) and more anxiety (N.S.Q.). Various comparisons within the malignant group suggest that women with cancers of non-sexual sites may be more neurotic and show less integration of the self-sentiment and that women with breast cancer may be characterized by more covert relative to overt evidence of anxiety. A species of item analysis, carried out using X 2, yielded 14 items which significantly differentiated the experimental from the control group. The bulk of these contribute towards anxiety but a few show the malignant group as less anxious. There is also some limited evidence of emotional toughness and depression on the part of the cancer patients. Women with lesions of non-sexual sites as distinct from sexual sites seem characterized by ‘ready admission’ and the higher anxiety score contributed by such a tendency. This is based on 4 items which significantly differentiate these 2 subgroupings. It seems best to interpret most of the data as consequent upon the subjects being cases of malignancy. The odd bits of data that might conceivably be construed as reflecting the pre-morbid personality of these women seem insufficient grounds upon which to stake a claim for a psychosomatic element in malignant disease. This may only mean that we have not yet hit upon the gauge which will give us the measure of something, upon which eminent men from Galen's time to the present have insisted—if not always with full accord as to its nature—was there. The very real possibility that cancer patients are given to motivational distortion in the context of self-rating instruments raises problems as to what form this gauge might take and still retain the properties of a simply administered screening device. In the last analysis it seems likely that future researches will have to select the more promising findings of such retrospective, metric researches as exist and enter upon a long-term, prospective study akin to that reported by Hagnell . Only in this fashion can the probability of there being a predisposing entity in malignant etiology which takes the form of some aspect of personality be reliably assessed. Meanwhile, inadequate as they must be with respect to a final answer, the smaller retrospective studies can contribute some sort of preliminary reconnaissance of a more than usually tangled but interesting battlefield.