Abstract In biofluid, long RNAs are more informative than microRNAs in terms of gene number and variation type. Therefore, cell-free long RNAs have shown promising potential as biomarkers in liquid biopsy, while they are mostly fragmented. In order to investigate these fragmented cell-free RNAs (cfRNAs), we developed a cost-effective cfRNA sequencing method, DETECTOR-seq (depletion-assisted multiplexing cell-free total RNA sequencing). It utilized a set of customized guide RNAs to remove large amounts of unwanted RNAs (i.e., fragmented ribosomal and mitochondrial RNAs) in human plasma. Early barcoding was also incorporated to save cost and plasma volume. After demonstrating its superior performance to other methods, we used DETECTOR-seq to investigate cell-free transcriptomes in whole human plasma and extracellular vesicles (EVs) it contains. We first observed different type distributions: structured circular RNAs, tRNAs, Y RNAs, and virus RNAs were enriched in plasma, while mRNAs and srpRNAs were enriched in EVs. We also uncovered distinct functional pathways: RNA splicing-related ribonucleoproteins (RNPs) and antimicrobial humoral response genes were enriched in plasma, while transcriptional activity, cell migration, and antigen receptor-mediated immune signals were enriched in EVs. Subsequently, we compared the performances of these distinct cfRNAs in whole plasma versus EVs on classifying cancer patients. The accuracies were comparable when discriminating cancer patients from healthy donors (AUCs: 0.936 versus 0.953). Meanwhile, cancer types (i.e., colorectal versus lung cancer) were better classified with microbial cfRNAs in plasma than in EV (AUCs: 0.898 versus 0.772). Overall, by investigating total and EV cfRNAs in the pairwise plasma samples, our work provides practical guidance for the proper decision of EV purification when launching a cfRNA-based study. Furthermore, as a cost-effective method, DETECTOR-seq would facilitate transcriptome-wide studies in the fields of extracellular RNA biology and clinical liquid biopsy. Key Points DETECTOR-seq enables efficient and specific depletion of sequences derived from fragmented ribosomal and mitochondrial RNAs in plasma. Distinct cfRNA signatures in whole plasma versus EVs were revealed. Both Plasma and EV cfRNAs were capable of distinguishing cancer patients from normal individuals. Microbial RNAs in Plasma cfRNAs enabled better classification of cancer types than EV cfRNAs.