To study whether colony growth in vitro reflects the prognosis of breast cancer patients, specimens from a total number of 138 patients with primary breast carcinomas were cultivated in the Courtenay-Mills soft agar method. The plating efficiency (PE) values were related to various clinical and histopathological parameters. No significant correlation was found between colony forming ability and menopausal status, histopathology, TNM-status or steroid hormone receptor status. The crude survival of the patients was not significantly correlated to the in vitro growth of the tumours; neither was there any difference in relapse-free survival between patients whose tumours failed to grow in vitro and those having growing tumours (PE greater than 0). A multivariate survival analysis of 115 patients with primary tumours without distant metastases revealed that the PE was not a significant independent prognostic indicator, as it gave no additional prognostic information above that of node and ER status. It is concluded that routine measurement of colony formation in vitro is not warranted in the management of breast cancer.