Suppression of genes involved in DNA repair, tumor suppression and detoxification through epigenetic modifications has been implicated in the etiology of cancer. As such analysis of promoter methylation patterns in genes frequently down regulated in breast cancer in non-cancerous subjects may serve as an indicator of breast cancer risk. CpG-island hypermethylation of single genes has been detected in cells isolated from nipple aspirate and ductal lavage, yet both isolation methods yield insufficient cells to complete an extensive analysis on any one donor sample. As an alternative we have turned to magnetic separation of human mammary epithelial cells from breast milk. Initial studies with these cells, which are detailed in chapter one, show that a breast milk sample provides sufficient epithelial cells to isolate high quality RNA for gene expression analyses or genomic DNA for methylation analysis of multiple genes. Using quantitative RT-PCR of RNA collected from these samples we detected differences in the mRNA levels for six genes known to be down regulated in breast cancers: BRCA1, p16, CDH1, TMS-1, GSTPi, and SFRP1. Additionally using methyl-specific PCR (MSP) we assayed for a small panel of genes frequently methylated in cancer and found them to be unmethylated in the few breast milk samples examined. However, given the small number of CpG sites which can be assayed by the MSP technique it is not surprising that methylation was not detected in disease-free subjects. With methods for collecting breast milk samples and processing them for genetic material established we turned to a more comprehensive study of DNA methylation in larger population of donors which is detailed in chapter two. Utilizing a highly sensitive and highly quantitative methylation analysis technique known as Pyrosequencing we examined age-related methylation patterns for RASSF1A, TMS-1, CDH1, SFRP1, GSTPi, and CRBP1 in genomic DNA purified from exfoliated epithelial cells magnetically enriched from breast milk (n=111) and whether the protective effects conveyed by early pregnancy could be partly due to decreases in DNA hypermethylation. Although firm answers about early pregnancy were inconclusive based on our sample pool, this body of work lays down a solid foundation for future studies.