Classification of preinvasive breast disease could be better founded using biologic markers, thereby increasing reproducibility. We studied 57 breast ductal and lobular in situ carcinomas by means of comparative genomic hybridization and correlated these findings with quantitative features such as the mean nuclear area, mitotic index (MI), apoptotic index (AI), and the presence or absence of necrosis. Loss of 8p and gains of 8q and 6q were associated, respectively, with a significantly higher MI and AI, whereas loss of 16q was associated with a lower MI and AI. A significantly higher number of alterations per case were seen in tumors with gains of 6q, 8q, and 17q and tumors with loss of 13q. Loss of 16q and gain of 17q correlated with the absence or presence of necrosis, respectively. Our data clearly demonstrate that distinct cytogenetic changes correlate with phenotypic changes, proliferation, and apoptosis. These data may be used to refine existing classification schemes.