MicroRNAs (miRNAs) in tumor-derived extracellular vesicles (tEVs) are important cancer biomarkers for cancer screening and early diagnosis. Multiplex detection of miRNAs in tEVs facilitates accurate diagnosis but remains a challenge. Herein, we propose an encoded fusion strategy to profile the miRNA signature in tEVs for pancreatic cancer diagnosis. A panel of encoded-targeted-fusion beads was fabricated for the selective recognition and fusion of tEVs, with the turn-on fluorescence signals of molecule beacons for miRNA quantification and barcode signals for miRNA identification using readily accessible flow cytometers. Using this strategy, six types of pancreatic-cancer-associated miRNAs can be profiled in tEVs from 2 μL plasma samples (n = 36) in an isolation-free and lysis-free manner with only 2 h of processing, offering a high accuracy (98%) to discriminate pancreatic cancer, pancreatitis, and healthy donors. This encoded fusion strategy exhibits great potential for multiplex profiling of miRNA in tEVs, offering new avenues for cancer diagnosis and screening.