MicroRNAs (miRNAs) are present in human plasma and known as a non-invasive biomarker for cancer detection. Our study was designed to identify plasma miRNAs specific for rheumatoid arthritis (RA) by a comprehensive array approach. We performed a systematic, array-based miRNA analysis on plasma samples from three RA patients and three healthy controls (HCs). Plasma miRNAs with more than four times change or with significant (P<0.05) change in expression, or detectable only in RA plasma, were confirmed with plasma from eight RA patients and eight HCs using real-time quantitative PCR. Consistently detectable miRNAs that were significantly different between RA patients and HCs were chosen for further validation with 102 RA patients and 104 HCs. The area under curves (AUC) were calculated after plotting the receiver operating characteristic (ROC) curves. To determine if these miRNAs are specific for RA, the concentrations of these miRNAs were analyzed in 24 patients with osteoarthritis (OA), and 11 patients with systemic lupus erythematosus (SLE). The array analysis and the subsequent confirmation in larger patient cohort identified significant alterations in plasma levels of seven miRNAs. The highest AUC was found for miR-125a-5p, followed in order by miR-24 and miR-26a. Multivariable logistic regression analysis showed that miR-24, miR-30a-5p, and miR-125a-5p were crucial factors for making detection model of RA and provided a formula for Estimated Probability of RA by plasma MiRNA (ePRAM), employing miR-24, miR-30a-5p and miR-125a-5p, which showed increased diagnostic accuracy (AUC: 0.89). The level of miR-24, miR-125a-5p, and ePRAM in OA and SLE patients were lower than that in RA. There was no significant difference in detection for anti-citrullinated protein antibody (ACPA)-positive and ACPA-negative RA patients. These results suggest that the plasma concentrations of miR-24 and miR-125a-5p, and ePRAM are potential diagnostic markers of RA even if patients were ACPA-negative.