In a minority of patients with a discrete breast lump the initial cytological assessment is either unsatisfactory or at variance with the results of other methods of diagnosis. A randomized comparison of repeat cytology and needle-core biopsy provided clinically useful information in 14 of 31 patients receiving repeat cytology and in 26 of 29 randomized to core biopsy. Nineteen patients had carcinoma: ten who received repeat cytology, which indicated malignancy in only three (diagnostic of malignancy in one, suspicious in two), while all nine patients who underwent core biopsy had a correct diagnosis (only suspicious of malignancy in one). The sensitivity for the definitive diagnosis of carcinoma on repeat cytology and core biopsy was 10 and 89 per cent respectively. Patients with a discrete breast lump and unclear cytology results require needle-core biopsy. This has more diagnostic value than repeat cytology.