BACKGROUND Promoter hypermethylation has emerged as a promising cancer biomarker. Currently, a large variety of quantitative and non-quantitative techniques is used to measure methylation in clinical specimens. Here we directly compared three commonly used methylation assays and assessed the influence of tissue fixation, target sequence location and the amount of DNA on their performance. METHODS We used Methylation-Specific PCR (MSP), Quantitative Multiplex MSP (QM-MSP) and Methylation-Specific Multiplex Ligation-dependent Probe Amplification (MS-MLPA) to compare methylation of CCND2, SCGB3A1, RARB and RASSF1 on DNA from 40 breast carcinomas. RESULTS A comparison between MSP and QM-MSP on the same samples showed a high discrepancy: 20% of tumors that showed no methylation in MSP gave >10% methylation in QM-MSP. In contrast, QM-MSP correlated strongly with MS-MLPA when targeting the same sequence in DNA from paraffin embedded as well as fresh frozen tissue. This correlation declined when target sequences were non-overlapping. In titration experiments, MSP and MS-MLPA performed robust with 10 ng of DNA, while QM-MSP was at least ten-fold more sensitive. CONCLUSIONS Despite the difference in molecular basis, QM-MSP and MS-MLPA showed moderate to strong correlations. In contrast, there was a poor concordance between either of these techniques and non-quantitative MSP. For biological samples with scarce DNA, QM-MSP is the method of choice.