Direct-acting antiviral (DAA) therapy for chronic hepatitis C virus (HCV) infection patients (CH) results in a sustained viral response (SVR) in over 95% of patients. However, hepatocellular carcinoma (HCC) occurs in 1–5% of patients who achieved an SVR after treatment with interferon. We attempted to develop a minimally invasive and highly reliable method of predicting the occurrence and recurrence of HCC in patients who achieved an SVR with DAA therapy. The exosomal miRNA expression patterns of 69 CH patients who underwent HCC curative treatment and 70 CH patients were assessed using microarray analysis. We identified a miRNA expression pattern characteristic of SVR-HCC by using machine learning. Twenty-five of 69 patients had HCC recurrence. The expression of four exosomal miRNAs predicted HCC recurrence with 85.3% accuracy. Fifteen of 70 patients had HCC occurrence. The expression of four exosomal miRNAs predicted the onset of HCC with 85.5% accuracy. The expression patterns of miR-4718, 642a-5p, 6826-3p, and 762 in exosomes were positively correlated with those in the liver, and downregulation of these miRNAs induced cell proliferation and prevented apoptosis in vitro. Aberrant expression of four miRNAs, which was used for prediction, was associated with HCC onset after HCV eradication. Expression patterns of exosomal miRNAs are a promising tool to predict SVR-HCC.