Résumé : Quantitative image cytometry was used to compare 18 parameters relating to ploidy, nuclear area, and chromatin texture to axillary lymph node status, tumor size, and histological grade for 34 infiltrating ductal carcinomas, each of which had been graded independently by each of six surgical pathologists. Zinc formalin-fixed, paraffinembedded tumors were assessed using the Elston and Ellis modification of the Bloom and Richardson histological grading scheme. When axillary lymph node-negative tumors were compared with those involving four or more nodes, % 2 c (diploid) cells, nuclear area, and eight of 12 chromatin texture parameters showed statistically significant differences. Carcinomas < 2 cm had more % 2 c (diploid) cells and fewer % > 4 c (hypertetraploid) cells than larger neoplasms. For tumors having nuclear pleomorphism score two versus those with score three, nuclear area, four of five parameters related to ploidy level, each of five parameters related to run-length matrix features and one of four co-occurrence matrix features showed significant differences. Nearly all of these cytometric parameters also showed significant differences for histological grade and mitotic count, which was strongly correlated with nuclear pleomorphism. In examining the cytometric parameters in relation to the interobserver reproducibility of histological grade and its components, the largest number of statistically significant parameters related to the nonreproducibility of nuclear pleomorphism. The findings indicate that as the grade of infiltrating ductal carcinomas increases, there are fewer % 2 c (diploid) cells and more % > 4 c (hypertetraploid) and % > or = 5 c (polyploid) cells. In addition, the cells of high grade tumors have larger nuclear areas and more small and large dense chromatin clumps, which increase in such number that they tend to join together. When compared with the cytometric parameters, nuclear pleomorphism is the most sensitive component of grade to nonreproducibility.