Approximately 70% of breast cancers contain cell populations with hyperdiploid (greater than G0/G1) DNA content; however, cells cultured from breast cancers have only diploid DNA contents and karyotypes. Mechanically dissociated cells rarely, if ever, grow in culture, while enzymatically dissociated cells do grow in most cases. To determine if cell dissociation techniques used to prepare cells for culture and other laboratory procedures select for cells with specific features, and if tumor cells are killed in the process, breast cancer cells obtained by mechanical dissociation and by enzymatic dissociation were examined for DNA content and cell viability (measured by dye exclusion). Mechanical dissociation yielded more dead cells and cells with hyperdiploid (greater than G0/G1) DNA than did enzymatic dissociation. Hyperdiploid cells were also found in the dye-excluding population with each dissociation technique, suggesting that the hyperdiploid cells were not always dead. We conclude that, in vivo, tumors contain cellular subpopulations with low viability and hyperdiploid (greater than G0/G1) DNA patterns. The extent to which these subpopulations are present in a sample depends on the dissociation technique employed. That only diploid cells are found in cultures of primary breast cancers may be because enzymatic dissociation, used to prepare cells for culture, yields predominantly diploid cells. These observations also have important implications for interpreting measurements made on dispersed cells, e.g., viability, DNA content, and other cytochemical markers.