This exploratory study is an attempt to define psychological attributes related to longer survival in patients with metastatic cancers. Previous published analyses have been limited in two ways. First, they have almost always been carried out on patients not receiving therapy; we have followed people receiving a year of group therapy, on the assumption that if mental qualities are to affect cancer progression, substantial mental change would be needed to alter the established balance between the cancer cells and host regulatory mechanisms. Second, the methods typically used to characterize patients' psychology have been self-report inventories, and many decades of research with such methods have largely failed to produce a consensus on what mental qualities, if any, promote survival. By contrast, we have used qualitative methods, allowing a much more in-depth analysis of the patients, without preliminary assumptions as to what would be important. The present report describes the results of a detailed qualitative analysis of data collected from 22 participants over a year of weekly group therapy. Using grounded methods, categories were derived from the extensive verbal data (comprising patients' written homework and therapists' notes), and linked in a model of change. By applying ratings to some of these categories, and combining these ratings, we derived a quantitative estimate of patients' "involvement in self-help." Rankings on degree of involvement corresponded quite closely with the quality of patients' experience and with their survival duration. There was a great range in degree of involvement, and various subgroupings could be discerned. Nine of the participants were classed as "highly involved," meaning that they devoted regular daily time, often several hours, to such self-help strategies as relaxation, mental imaging, meditation, cognitive monitoring and journalling. All but 1 of these patients enjoyed a good quality of life and lived at least 2 years. Two of them have remained in complete remission for 7 years. At the other end of the scale, 8 patients showed little application to the work, being either unconvinced that it would help them or hampered by psychological problems such as low self-esteem. None of these was rated as having a good quality of life, and only 1 lived more than 2 years, although, as a group, their medical prognoses were no more unfavorable at the onset of therapy than for the "high involvement" group. The different subgroups and aspects of the model are illustrated by representative quotations.