In order to test the effect of a psychological intervention on survival from cancer, 66 women with metastatic breast cancer, all receiving standard medical care, were randomly assigned into two groups; one group (n = 30) attended the psychological intervention, consisting of 35 weekly, 2 h sessions of supportive plus cognitive behavioral therapy; the control group (n = 36) received only a home study cognitive behavioral package. No significant difference was found in survival post-randomization between the groups as assessed by a log rank test 5 years after the commencement of the study. As expected, several prognostic factors were significant predictors of survival: metastatic site, hormonal receptor status, and chemotherapy prior to randomization. While many personal and demographic variables did not influence survival, there was a significant effect of self-reported exercise (possibly due to better health). A small subgroup of intervention subjects who attended outside support groups also survived significantly longer than those who did not. The strengths and limitations of the present study are discussed, and the results contrasted with those of a well known study by Spiegel et al. (Spiegel, D., Bloom, J.R., Kraemer, H.C. and Gottheil, E. (1989) Lancet ii, 888-891). We propose that a different experimental design (correlative) may be needed to show any effect of self-help behaviors and psychological attributes in a small minority of patients.