BACKGROUND Circulating microRNAs (miRNAs) have recently been proposed to be biomarkers for various diseases including cancer and cardiovascular disease. Erythrocytes are a major source of miRNAs in blood. However, it remains unknown how miRNA levels in serum are influenced by miRNAs in erythrocytes. In this study, we investigated the relationships among serum levels of miRNAs that are contained in erythrocytes. METHODS Participants were middle-aged healthy Japanese men. Total miRNAs in serum from each participant were analyzed using the 3D-Gene miRNA Oligo chip. Relationships among the levels of eleven miRNAs (miR-103a-3p, -144-3p, -15a-5p, -16--5p, -26a-5p, -423-5p, -451a, -484, -486-5p, -92a-3p, and -93-5p) that have been reported to exist in erythrocytes were investigated by using correlation analysis. RESULTS Among 55 pairs prepared by the above 11 miRNAs, there were significant correlations between miRNA levels of 31 pairs. In principal component analysis, 4 major erythrocyte-derived miRNAs, miR-16-5p, -451a, -486-5p and -92a-3p, were included in the first principal component. There were strong correlations between miR-16-5p and -451a levels (Spearman's rank correlation coefficient: 0.920) and between miR-486-5p and -92a-3p levels (Spearman's rank correlation coefficient: 0.863). CONCLUSIONS There are significant associations among serum levels of erythrocyte-derived miRNAs, and these associations should be taken into account when considering the miRNAs as disease biomarkers.