Multimodal characterization of cell-free DNA (cfDNA) in blood can enable the sensitive and non-invasive detection of human cancers but remains technically challenging and costly. Here, we developed Multimodal Epigenetic Sequencing Analysis (MESA), a flexible and sensitive method of capturing and integrating multimodal epigenetic information of cfDNA using a single experimental assay, i.e., non-disruptive bisulfite-free methylation sequencing, such as Enzymatic Methyl-seq (EM-seq) and TET-assisted pyridine borane sequencing (TAPS). MESA can simultaneously infer four epigenetic modalities, namely cfDNA methylation, nucleosome occupancy, nucleosome fuzziness, and fragmentation profile for regions surrounding gene promoters and polyadenylation sites (PASs). When applied to 462 cfDNA samples from 2 independent clinical cohorts for colon cancer, new modalities (e.g., nucleosome fuzziness) and genomic features (e.g., PASs) introduced in MESA are highly complementary or superior to conventional ones, such as promoter DNA methylation, for cancer detection. Furthermore, MESA’s integrated analysis of multimodal epigenetic features significantly improved the detection accuracy for colon, liver, and pancreatic cancers compared to single modality models. Together, MESA captures additional and highly complementary epigenetic information from cfDNA without additional experimental assays, highlighting the importance and clinical potential of using multimodal epigenetic features for non-invasive cancer detection.