Malignant pleural mesothelioma (MPM) is a rare and aggressive cancer mainly caused by asbestos exposure. Specific and sensitive non-invasive biomarkers may facilitate and enhance screening programs for the early detection of cancer. We investigated DNA methylation (DNAm) profiles in MPM pre-diagnostic blood samples in a case-control study nested in the European Prospective Investigation into Cancer and nutrition (EPIC) cohort, aiming to characterise DNAm biomarkers associated with MPM. From the EPIC cohort, we included samples from 135 participants who developed MPM during 20 years of follow up and from 135 matched, cancer-free, controls. For the discovery phase we selected EPIC participants who developed MPM within five years from enrolment (n=36) with matched controls. We identified nine differentially methylated CpGs, selected by 10-fold cross-validation and correlation analyses: cg25755428 (MRI1), cg20389709 (KLF11), cg23870316, cg13862711 (LHX6), cg06417478 (HOOK2), cg00667948, cg01879420 (AMD1), cg25317025 (RPL17) and cg06205333 (RAP1A). Receiver operating characteristic (ROC) analysis showed that the model including baseline characteristics (age, sex, PC1wbc) along with the nine MPM-related CpGs has a better predictive value for MPM occurrence than the baseline model alone, maintaining some performance also at more than five years before diagnosis [AUC (area under the curve) < 5 years=0.89; AUC 5-10 years=0.80; AUC >10 years=0.75; baseline AUC range=0.63-0.67)]. DNAm changes as non-invasive biomarkers in pre-diagnostic blood samples of MPM cases were investigated for the first time. Their application can improve the identification of asbestos-exposed individuals at higher MPM risk in order to possibly adopt more intensive monitoring for early disease identification.