Identifying disease-related variants is a primary aim of human genetics. In breast cancer, loss of heterozygosity at specific loci was previously demonstrated in paired tumour and circulating plasma cell-free DNA (cfDNA) samples. However, alterations unique to cfDNA were also found in all cases, suggesting disease progression. These results prompted the characterisation of the circulating breast cancer genome in more detail in this thesis, to test the hypothesis that cfDNA acts as a surrogate tumour marker. This was achieved using Affymetrix SNP 6.0 technology and bioinformatics to map SNP and copy number variation (CNV), comparing cfDNA with matched normal lymphocyte and tumour DNA in 65 breast cancer patients and 8 healthy female controls. Results in this thesis show that comparison of cfDNA SNP genotypes can distinguish between primary breast cancer patients and healthy controls (p<0.0001), and between pre-surgical breast cancer patients and those who already had surgery/treatment (p=0.0016). A significant difference (p=0.0006) was also found between cfDNA samples taken an average of 3 years apart in women on follow-up, again suggesting progression. In addition, CNV amplification was observed in matched tumour and cfDNA at numerous loci on different chromosome arms. Many of these tumour-specific CNVs contributed significantly to disease through logistic regression analysis and remained detectable in cfDNA up to 12 years after diagnosis despite no other evidence of disease. This finding strongly infers breast cancer dormancy in the majority of patients on follow-up. In addition, candidate CNVs were validated by real-time qPCR and additional bioinformatics revealed key SNP and CNV signatures of breast cancer patients. If validated in other patient series, the results could alter the diagnostic and prognostic landscape of breast cancer. Future studies will focus on developing high throughput approaches to target common SNPs/CNVs with a view to developing a targeted custom DNA chip for screening and monitoring.