The present study describes the setting up of a new score which makes it possible objectively to grade ductal breast carcinomas, i.e. not-otherwise-specified (NOS) cancers, on cellular material from fine-needle aspirations (FNAs). For this purpose FNAs from 252 patients--with NOS breast cancers--were smeared onto histological slides, fixed in an ethanol-formolacetic acid mixture, Feulgen-stained and analysed by means of a cell image processor. Four parameters were taken into account in setting up the score, namely the cytological prognostic grade (CPM) of malignancies similar to the Scarff-Bloom-Richardson (SBR) grading, the nuclear area (NA), the DNA index (DI) and the DNA histogram type (DHT). Each of these four parameters was considered as a "sub-score" which may take three values, i.e. 1, 2 and 3. The final result may thus range from 4 to 12. Subscores of 4 and 5 correspond to a cytological score of I, subscores of 6, 7 and 8 to a cytological score of II, sub-scores of 9 and 10 to a cytological score of III, and sub-scores of 11 and 12 to a cytological score of IV. In the present study, the results show 17% of CPM grade 1.52% of CPM grade II and 31% of CPM grade III cancers. All the cases exhibiting a cytological score of IV (5%) fully fit in with the CPM grade III cancers. In the same way, none of the cases exhibiting a score of I fit in with CPM grade III cancers. The cancers with a CPM grade II fit in with the scores of II and III. It thus seems possible to convert a three-value malignancy grading system (CPM and/or SBR grading) into a four-value one (cytological score). The main advantage in this latter type of system is that it becomes possible to split up the over-large group of CPM grade II cancers. As things stand, we are unable to give any prognostic value for the score proposed here because our study is prospective only. A study of this type has been necessary so as to provide against problems connected with ways of preserving specimens that might be used in a retrospective study. The bank of clinical and biological data now in existence must be allowed to mature for a number of years before the prognostic worth of the cytological score can be established, always assuming that such a value exists.