Employing reference genes to normalize the data generated with quantitative PCR (qPCR) can increase the accuracy and reliability of this method. Previous results have shown that no single housekeeping gene can be universally applied to all experiments. Thus, the identification of a suitable reference gene represents a critical step of any qPCR analysis. Setaria viridis has recently been proposed as a model system for the study of Panicoid grasses, a crop family of major agronomic importance. Therefore, this paper aims to identify suitable S. viridis reference genes that can enhance the analysis of gene expression in this novel model plant. The first aim of this study was the identification of a suitable RNA extraction method that could retrieve a high quality and yield of RNA. After this, two distinct algorithms were used to assess the gene expression of fifteen different candidate genes in eighteen different samples, which were divided into two major datasets, the developmental and the leaf gradient. The best-ranked pair of reference genes from the developmental dataset included genes that encoded a phosphoglucomutase and a folylpolyglutamate synthase; genes that encoded a cullin and the same phosphoglucomutase as above were the most stable genes in the leaf gradient dataset. Additionally, the expression pattern of two target genes, a SvAP3/PI MADS-box transcription factor and the carbon-fixation enzyme PEPC, were assessed to illustrate the reliability of the chosen reference genes. This study has shown that novel reference genes may perform better than traditional housekeeping genes, a phenomenon which has been previously reported. These results illustrate the importance of carefully validating reference gene candidates for each experimental set before employing them as universal standards. Additionally, the robustness of the expression of the target genes may increase the utility of S. viridis as a model for Panicoid grasses.