Cancer-derived exosomes carry a large number of specific molecular profiles from cancer cells and have emerged as ideal biomarkers for early cancer diagnosis. Accurate detection of ultralow-abundance exosomes in complex biological samples remains a great challenge. Herein, a novel SERS aptasensor powered by cascaded signal amplification of CRISPR/Cas13a trans-cleavage and catalytic hairpin assembly (CHA) was proposed for ultrasensitive detection of gastric cancer-derived exosomes, which included hairpinstructured recognition aptamers (MUC1-apt), cascaded signal amplification (i.e., CRISPR/Cas13a trans-cleavage and CHA), SERS tags, and silver nanorods (AgNRs) sensing chip. In the presence of SGC-7901 cell-derived exosomes, MUC1-apt specifically bound to MUC1 proteins highly expressed on exosomes via its contained MUC1 aptamer with its exposed RNA fragments activating the CRISPR/Cas13a trans-cleavage to cleave the uracilmodified hairpin reporter, and the cleavage products further triggered the downstream CHA reaction to form numerous duplexes, which can, in turn, capture a large number of SERS tags onto the AgNRs sensing chip to generate a significantly enhanced Raman signal. The proposed SERS aptasensor exhibits good performance on analysis of exosomes, i.e., rapid response within 60 min, singleparticle sensitive detection from a 2 μL biological sample, good specificity in distinguishing SGC-7901 cell-derived exosomes against other exosomes, good uniformity, excellent repeatability, and satisfactory recoveries in human serum, and good universality to expand the detection of multiplex exosomes, which indicates that the SERS aptasensor provides a valuable reference for clinical diagnosis of early cancer.