Chromosome analysis was performed on samples from 85 consecutive patients with breast cancer by one or more of three different methods: direct harvest, culture after mechanical disaggregation, and culture after collagenase digestion. Metaphases suitable for karyotyping were obtained in 70% of the cases; direct harvest yielded metaphases in 29% and cultures without and with digestion in 40% and 59%, respectively. Chromosomal abnormalities were detected in 37 cases. Cells judged to be phenotypically abnormal in culture were twice as likely to reveal chromosomal aberrations as normal-looking cells. Eight cases showed multiclonal abnormalities. Significant differences were detected in the karyotypic profile depending on the method used. With direct harvest, the yield of complex chromosomal changes was 87%, compared to 44% after culture of digested tissue (P < 0.01), and also polyploidy was more common in direct-harvested samples. Detailed karyotypic analysis was possible in 29 primary tumors. The chromosomes most frequently involved were 1, 3, 7, 11, 16, and 17. Recurrent structural abnormalities were der(1;16)(q10;p10), i(1)(q10), del(6)(q21), and del(1)(p22). Breakpoints clustered to the centromere regions of chromosomes 1, 3, 11, 15, and 16 and to the short arms of chromosomes 7, 17, and 19. Seven of twenty-nine fully analyzed cases had a family history of breast cancer, and changes of chromosomes 1, 3, and 15 seemed to be more common in these cases. There was an association between karyotype and survival: The 3 year survival was 63% in patients with complex karyotypic changes and 92% in those without complex changes.