We analysed 1500 consecutive fine needle aspiration cytology (FNAC) specimens to ascertain the reasons for diagnostic failure. Of 221 tumours proven malignant following open biopsy, 184 (83%) were correctly diagnosed on FNAC. Of 1082 aspirates classified 'benign', 787 (73%) underwent open biopsy and of these 33 (4%) were diagnosed malignant. All three cases of ductal carcinoma in-situ (DCIS) and eight or 16 lobular carcinomas in this study were missed using FNAC alone. Twelve of the 22 patients with invasive carcinoma not diagnosed oncytology had tumours measuring less than 1 cm diameter. Six of 1500 FNAC reports (0.4%) gave false positive diagnoses, rive were classified as 'suspicious of malignancy; and one as frankly malignant. The overall sensitivity was 84%, specificity 99% and positive predictive value 97%. Though these results confirm the value of FNAC as a rapid means of diagnosing most breast cancers, it it unreliable in patients with invasive carcinomas less than 1 cm in diameter and for the detection of lobular and in-situ carcinoma.