Aberrant DNA methylation is ubiquitous in human cancer and has been shown to occur early during carcinogenesis, thus providing attractive potential biomarkers for the early detection of cancer. The introduction of genome-wide DNA methylation analysis comparing tumor and nonmalignant tissues resulted in the discovery of many regions that undergo aberrant methylation during carcinogenesis. Those regions can potentially be used as biomarkers for cancer detection. However, a biomarker will be useful for screening or early detection of cancer only if it can be detected in a noninvasive or minimally invasive fashion without tissue biopsy. The authors discuss the challenges in translating DNA methylation biomarkers to cancer diagnosis - including obstacles in assay development, tissue-specific methylation load on tumor suppressor genes, detecting markers with sufficient sensitivity and specificity in the periphery, and ways in which these obstacles can be overcome.