Despite the promising value of miRNAs in the diagnostic and prognostic of cardiovascular disease (CVD), recent meta-analyses did not support their potential. Methodological variances in studies may interfere with miRNA profile and affect their results. This study determines if the blood starting material is a source of variance in miRNA profile by performing a paired comparison in plasma and serum of the expression of primary miRNAs associated with CVD. Circulating miRNA yield was similar in both plasma and serum, although a significant increase was observed in patients with Non-ST-elevation myocardial infarction (NSTEMI) compared to control volunteers. When normalized by the expression of miR-484, different patterns of miRNA expression between serum and plasma. Although NSTEMI modified the expression of miR-1 and miR-208 in both serum and plasma, plasma displayed a higher variance than serum (Levene's test p < 0.01). For miR-133a and miR-26a, differences were only detected in serum (p = 0.0240), and conversely, miR-499a showed differences only in plasma of NSTEMI (p = 0.001). Interestingly, miR-21 showed an opposite pattern of expression, being increased in serum (2-ΔΔCt: 5.7, p = 0.0221) and decreased in plasma (2-ΔΔCt: 0.5, p = 0.0107). Plasma and serum exhibit different patterns of circulating miRNA expression in NSTEMI and suggest that results from studies with different starting material could not be comparable.