Tumours from 117 patients with breast cancer operated during a 5-month period 1975-1976 were investigated by absorbance scanning cytophotometry. The relations of these data to breast cancer recurrence and death during a follow-up period of 8.5 years were analysed using Cox's proportional hazards model. When individually tested, nodal status, grade of malignancy and high rates of proliferation, as indicated by cells in S-phase, were statistically significant predictors of the clinical outcome. The risk ratio associated with S-phase decreased significantly over time from primary treatment. No significant relation was found between DNA-ploidy or tumour size and the clinical course. In the multivariate analysis, based on 76 patients, nodal status alone was a significant prognostic factor.