MicroRNAs (miRNAs) are highly conserved small non-coding RNAs, that modulate gene expression by targeting messenger RNA of many processes. Thus, miRNAs are key regulators of both physiological and pathological settings. Reliable results of quantitative miRNA evaluation depend on suitable reference genes (RGs) for data normalization. To date, no consensus has been reached on the best RG for muscle tissue. We assessed RGs stability in skeletal muscle tissue in patients with spinal deformity. Ninety tissue samples were obtained from the deep paravertebral muscles from the convex and concave sides of the spinal curvature, as well as the superficial paraspinal muscles. We evaluated the stability of twelve miRNAs (hsa-miR-1–3p, hsa-miR-1–5p, hsa-miR-26b-5p, hsa-miR-92a-3p, hsa-miR-133a-3p, hsa-miR-133a-5p, hsa-miR-133b, hsa-miR-191–5p, hsa-miR-206, hsa-miR-208b-5p, hsa-miR-486–5p, hsa-miR-499a-5p), finding three to be indicative of reference miRNA, and nine as muscle-tissue specific. Stability was quantified using four statistical tools and a comprehensive ranking system. Three miRNAs were indicated as the most stable, and we assessed hsa-miR-486–5p as the most, and hsa-miR-208b-5p as the least suitable RGs for miRNA quantitative analyses. We recommend using a minimum of three RGs miRNA to normalize RT-qPCR data. Finally, qPCR efficiency should always be considered. To obtain consistent results, data normalization in muscle tissue is required.