A better understanding of secondary metabolites biosynthesis requires comprehensive research at the molecular level. Although the medicinal importance of secondary metabolites extracted from Arnica spp. has been well documented, the very plants themselves have been poorly studied. Characterization of biosynthetic pathways and their complex regulatory mechanisms may be enhanced via transcriptomic approach. To ensure the reliability of data, RT-qPCR based gene expression studies should be preceded by the reference genes selection step. Here, ten candidate reference genes were analyzed in terms of their expression stability in developing flowers of two arnica species, Arnica montana and Arnica chamissonis. After evaluating their expression stability with four distinct algorithms (geNorm, NormFinder, BestKeeper and delta Ct method), we found that SKIP16 and F-box were the most stable reference genes in A. montana, whereas SAND and F-box were best-performing genes in A. chamissonis. On the other hand, ACT1 in combination with F-box could be used as internal control for cross-species investigations. This study will aid in further research on molecular mechanisms underlying the processes of flowering and secondary metabolites production in medicinal plants.