General immune competence was measured before treatment in 185 breast cancer patients. They were then followed for 5 to 11 years to determine its relationship to recurrence and its clinical value in predicting prognosis. The tests of immune competence used were immunoglobulins IgG, IgA, IgM, leucocyte counts, percentage and total lymphocyte counts and Mantoux and DNCB skin hypersensitivity tests.None of these tests was strongly predictive of recurrence on an individual basis, a finding similar to our results at 2 years. The longer period of follow-up now reported has provided no findings of unequivocal statistical significance, but suggests a biphasic host response to early tumours. The patients who developed recurrence within 5-11 years due to micrometastasis had higher lymphocyte counts in their preoperative assessment than patients who remained recurrence free. This suggests that small tumour volumes do not stimulate immunity and that large volumes depress it; turnouts in between these groups are associated with higher levels. Examination of studies by a number of authors reveal parallel findings which have not been previously noted. It is not possible to confirm the significance of these findings from this study because of the heterogeneity of human breast cancer. However, if they indicate a general principle of a dynamic host-tumour interplay they have important implications for assessing immmune competence at any single point of time and for the theory that cancer may arise during an anergic state. We hope that these findings will stimulate other workers to examine host-tumour interaction from this point of view.