[Truncated abstract] Introduction: Breast cancer is a heterogeneous disease, resulting in very different outcomes for women with apparently similar tumour characteristics. In order for patients to have optimal treatment, a better understanding of the molecular nature of their disease is required. Aims: The aims of this thesis were: 1) To determine whether methylation of RARβ2, ER, CDH1, BRCA1, CCND2, p16 and TWIST genes are associated with phenotypic features of breast cancer and the prognostic significance of methylation of these genes. 2) To investigate for possible associations between the frequency of methylation at RARβ2, CDH1, ER, BRCA1, CCND2, p16 and TWIST genes and the presence of germ-line variants in the TS, MTHFR, MS, CBS, MTHFD1 and DNMT3B genes, as well as for possible correlations between these polymorphisms and clincopathological features of breast cancer including patient outcome. 3) To determine whether PIK3CA mutations determined clinical phenotype and the prognostic significance of PIK3CA mutations in a large and well characterized cohort of breast cancer patients. Methods: A large and well characterized series of primary breast tumours were selected for methylation of RARβ2, ER, CDH1, BRCA1, CCND2, p16 and TWIST genes using MSP, and for polymorphisms in TS, MTHFR, MS, CBS, MTHFD1 and DNMT3B genes using PCR, PCR-RFLP and PCR-SSCP. Mutations to PIK3CA were detected using F-SSCP. Results and Conclusions: Methylation frequencies ranged from 11% for CCND2 to 84% for ER. More frequent hypermethylation was observed in tumours with poor histological differentiation compared to those with well/moderate differentiation, as well as trends for association with larger tumour size and mutant TP53. Tumours with ER and CDH1 methylation were associated with significantly lower hormone receptor levels, younger age at diagnosis and the presence of mutant p53. TWIST methylation is firstly reported to be associated with significantly older patient age at diagnosis and larger tumour size. Our data suggests that gene methylation may be linked to various pathological features of breast cancer. However, there appears to be little support for a distinctive CpG island methylator phenotype in breast cancer.