Fine-needle aspiration cytology (FNAC) is a widely practiced technique in the diagnosis of breast carcinoma, and it is the only diagnostic procedure performed before definitive treatment, at most institutions. While the histological grading of breast carcinoma has become routine in many centers worldwide, the cytopathological grading of breast carcinoma is not commonly used. Grading of breast carcinoma, while the tumor is still in vivo, would be the most ideal and desirable situation, as it would be helpful in the selection of patients for appropriate therapy. The objective of this study, therefore, was to devise a simple system for grading breast carcinoma, based on the cytological features alone. We reviewed 125 cases of breast carcinoma retrospectively, which were initially diagnosed by FNAC, with subsequent histopathological confirmation. These included 105 ductal, 6 lobular, 2 tubular, 1 papillary, and 1 medullary carcinoma. There was 1 ductal carcinoma in situ. Nine cases were rendered insufficient for grading. Thus 105 cases of ductal carcinoma (NOS) were evaluated for final cytological grading. Air-dried Diff-Quik-stained smears were reviewed at least twice independently by four histopathologists and were then compared with the original histological grades. Six cytological features used for grading were found to be statistically significant: cellular pleomorphism, nuclear size, nuclear margin, nucleoli, naked tumor nuclei, and mitoses. A scoring system based on these six essential parameters was used, to classify ductal carcinoma into three cytological grades, which showed close correlation with the established histological grades. In addition, two less consistent, but still important, features were the presence or absence of necrosis and stromal invasion. Another six parameters, including smear cellularity, degree of cell dispersion or clustering, lymphoplasmacytic infiltrate, presence of tubular structures, cytoplasmic appearance of the tumor cells, and smear background, were not statistically significant. However, these additional parameters were found helpful in assigning the correct grade, in cases with borderline scores. The concordance rate with histology was 100% for grade 1, 98% for grade 2, and 93% for grade 3. Diagn. Cytopathol. 2003;29:185–193. © 2003 Wiley-Liss, Inc.