Breast cancer is considered to display a high degree of intratumor heterogeneity, without any obvious morphological and pathological steps to define sequential evolution, and its progression may vary among individual tumors. In an attempt to elucidate these etiological and phenotypic complexities, the present study, based on the fundamental concept that genomic instability is the engine of both tumor progression and tumor heterogeneity, was conducted to test the hypothesis that breast cancer pathogenesis is driven by double-strand break (DSB)-initiated chromosome instability (CIN). The rationale underlying this hypothesis is derived from the clues provided by family breast cancer syndromes, in which susceptibility genes, including p53, ATM, BRCA1 and BRCA2, are involved within the common functional pathway of DSB-related checkpoint/ repair. Because genomic deletion caused by DSB is reflected in the genetic mechanism of loss of heterozygosity (LOH), this genome-wide LOH study was conducted, using 100 tumors and 400 microsatellite markers. To minimize the effect of heterogeneity within tumors, the experimental technique of laser capture microdissection was used to ensure that genetic and phenotypic examinations were based on the same tumor cells. Support for our hypothesis comes from the observations that: (a) the extent of DSB-initiated CIN in tumors significantly increased as tumors progressed to poorer grades or later stages; (b) in the sequential steps toward CIN, the loci of p53 and ATM, the key checkpoint genes against DSB, were lost at the earliest stage; and (c) many loci identified to be important in breast tumorigenesis were the genomic sites possibly harboring the genes involved in DSB-related checkpoint/repair (including RAD51, RAD52, and BRCA1) or CIN (including FA-A, FA-D, and WRN), and a higher number of these loci showing LOH was significantly associated with increased level of DSB-initiated CIN (P < 0.0001). Breast cancers are thus considered to be sequentially progressive with CIN. However, CIN might also cause genetic heterogeneity, which was revealed by the findings that LOH at some markers was observed only in the component of ductal carcinoma in situ but not in the invasive component of the same tumors. In addition, some markers were found to preferentially lose at specific tumor grades, implying their contribution to genetic heterogeneity during tumor development. Therefore, this study suggests that breast cancer progression is clonal with regard to CIN, but different breast cancers would present distinct molecular profiles resulting from genetic heterogeneity caused by CIN.