The clinical value and limitations of a previously described cytomorphometric method, based on nuclear size and anisonucleosis, for evaluation of routine fine-needle aspiration of the breast were assessed in a series of 313 histologically investigated primary breast lesions. Limitations were analyzed by histomorphometry and DNA flow cytometry in 116 consecutive cases of histologically confirmed breast carcinoma. Eighty per cent of histologically proven malignant tumours were classified cytomorphometrically as malignant and no false-positive results were encountered. For benign lesions a benign cytomorphometric classification was reached in 66% of the cases. Histomorphometry showed that on the whole the assignment of histologically malignant tumours to a non-malignant cytomorphometric classification was determined by smaller nuclei and not by sampling error. Tumours assigned to a malignant cytomorphometric classification had on average significantly higher DNA indices than did tumours not assigned to a malignant cytomorphometric classification (P less than 0.001). The mean-nuclear areas in cytomorphometry and histomorphometry were strongly correlated with DNA indices indicated by DNA flow cytometry (P less than 0.001 for both). The present findings show that this cytomorphometric method is appropriate for routine quality control of a cytological diagnosis of malignancy in FNA of the breast. However, an inconclusive result in 15-25% of the tumours is inevitable.