A series of seven studies investigating biopsychosocial aspects of cutaneous malignant melanoma were conducted by a multidisplinary group of researchers at the University of California, San Francisco. Two studies investigated the relationship of variables derived from a videotaped psychosocial interview and from self-report measures, and two histopathologic indicators: tumor thickness and level of invasion. Associations of psychosocial variables to prognostic indicators were stronger within the younger vs the older subject group. In a multiple regression analysis, patient delay in seeking medical attention emerged as the most significant variable predicting tumor thickness. A study of factors related to patient delay found longer delays in patients who had lesions on the back, less previous knowledge of melanoma, less understanding of its treatment and less minimization of its seriousness. Another study compared the repressive coping reactions—defined as the discrepancy between reported anxiety and that reflected in electrodermal activity—in melanoma patients, cardiovascular disease patients and disease-free controls. The melanoma group was significantly more ‘repressed’ on the combined self-report/physiological measure, as well as on other self-report measures of repressiveness. In order to investigate the relationship of psychosocial factors to more disease-relevant physiological variables, the next study focused on two clinical variables significantly predictive of disease outcome: mitotic rate of the tumor and lymphocytes at tumor site. Emotional expression of sadness and anger, rated from the videotaped interviews, was positively correlated with tumor-specific host-response factors and negatively correlated with mitotic rate. In another study, subjects who had died or had disease progression were matched on the basis of tumor and demographic characteristics with subjects who had no evidence of disease by follow-up. The unfavorable outcome group had higher scores on self-report scales of dysophoric emotion and distress which were administered one to three years previously. An experimental investigation of relationships among behavioral, physiological and tumor outcome variables in the Syrian hamster found that general activity was correlated positively with greater tumor growth following induction. These results were compared to those from the preceeding two human studies, and discussed in terms of a stress-arousal-coping model.