Objective To analyze the differential expression levels of plasma microRNAs (miRNAs) in prostate cancer patients, and to investigate the potential of combined plasma miRNA detection in the diagnosis of prostate cancer. Methods The Gene Expression Omnibus database was used to screen out the differentially expressed miRNAs in prostate cancer plasma or serum, and 3 miRNAs with the highest weights were screened out by using nearest centroid classifier for area under curve optimization (NCC-AUC) method. A total of 45 plasma samples from patients with prostate cancer (prostate cancer group), 40 from patients with chronic prostatitis (prostatitis group) and 42 from healthy physical examination participants (healthy control group) were collected from Shanghai Baoshan Hospital of Integrated Traditional Chinese and Western Medicine. The expression levels of candidate plasma miRNAs were verified by quantitative polymerase chain reaction. The diagnostic value of candidate miRNAs and combined miRNAs for prostate cancer was analyzed by receiver operating characteristic (ROC) curve. Results A total of 26 differentially expressed miRNAs related to prostate cancer were obtained by bioinformatics analysis, and 3 miRNAs (miRNA-21-5p, miRNA-5189-5p, and miRNA-6780a-5p) with the highest weights in NCC-AUC mode were included for the clinical validation stage. Clinical verification results showed that the expression level of plasma miRNA-21-5p was increased successively in the healthy control group, prostatitis group and prostate cancer group, and the differences between the 3 groups were statistically significant (P<0.01). The expression level of plasma miRNA-5189-5p in the prostatitis group was significantly lower than that in the prostate cancer group and healthy control group (both P<0.01). The expression level of plasma miRNA-6780a-5p in the prostate cancer group was significantly higher than that in the prostatitis group and healthy control group (both P<0.01). ROC curve showed that both plasma miRNA-21-5p and miRNA-6780a-5p had excellent diagnostic potential for prostate cancer (both P<0.001); the values of area under curve (AUC) were 0.695 and 0.787, sensitivities were 95.6% and 64.4%, and specificities were 41.5% and 81.7%, respectively. The combination of miRNA-21-5p and miRNA-6780a-5p could improve the diagnostic efficiency, with an AUC of 0.830, a sensitivity of 82.2%, and a specificity of 74.4%. Conclusion Plasma miRNA-21-5p and miRNA-6780a-5p are both potential diagnostic markers for prostate cancer, and the combined detection of both can improve the diagnostic efficacy.