Background: While circulating tumor DNA (ctDNA)-based assays for multi-cancer early detection (MCED) have shown significant potential, their performance is hindered by the limited amount and inherent variability of ctDNA, particularly in cancer types with low ctDNA shedding. To enhance the accuracy of MCED, recent studies have focused on profiling cell-free RNA (cfRNA), which is secreted not only by tumor cells but also by other cell types to reflect the systemic tumor responses. Herein, we developed a novel cell-free multi-omics approach integrating cfDNA and cfRNA analyses from a single blood draw to improve the sensitivity of an MCED assay, especially for detecting low-shedding ctDNA cancer types. Methods: We recruited 535 healthy subjects and 287 patients across five cancer types (77 breast cancers, 102 colorectal cancers, 35 gastric cancers, 34 liver cancers, and 42 lung cancers). We established a multi-omics workflow to comprehensively profile cfDNA signatures (methylation, fragment length, motif ends, and copy number alterations), and cfRNA signatures (cfmRNA). Distinctive signatures were identified to differentiate cancers from healthy subjects and used for constructing robust machine learning classifiers. Results: We identified multiple significantly different features distinguishing cancer from healthy individuals, with 81% originating from cfDNA. Notably, cfmRNA features, accounting for 19% of total significant features, were mainly enriched in immune-related pathways, reflecting tumor-immune cell interactions. The cfDNA-based model achieved 81.2% accuracy, with highest performance in detecting liver cancer (100%) and 64%, 42%, and 36.4% for breast, colorectal, and gastric cancers, respectively, at 93% specificity. Combining cfDNA and cfRNA signatures increased sensitivity to 73%, 71%, and 46% for breast, colorectal, and gastric cancers. Overall, the multi-omic model achieved 85.6% accuracy and 72.7% sensitivity for detecting five cancer types at 92.6% specificity. Conclusions: The cell-free multi-omics approach holds potential for improving ctDNA-based MCED tests by simultaneously profiling both cfDNA and transcriptomic cfmRNA from a single blood draw. Validation on larger cohorts could lead to a paradigm shift in cancer screening practices.