Real-time quantitative RT-PCR (qRT-PCR) is the most commonly used method for accurately detecting gene expression patterns. As part of qRT-PCR analysis, normalization of the data requires internal control gene(s) that display uniform expression under different biological conditions. However, no invariable internal control gene exists, and therefore more than one reference gene is needed to normalize RT-PCR results. In this study, we assessed the expression of eight candidate internal control genes, namely 18S ribosomal RNA (18S rRNA), elongation factor-1alpha, β-Actin, E2 ubiquitin-conjugating enzyme, β-Tubulin (TUB), ACTIN2, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and Msc27 of unknown function, in a diverse set of 16 alfalfa (Medicago sativa) samples representing different tissues and abiotic stress challenges, using geNorm and BestKeeper software. The results revealed that the eight candidate genes are inconsistently expressed under different experimental conditions. Msc27 and 18S rRNA are suitable reference genes for comparing different tissue types. Under different abscisic acid and NaCl conditions, three reference genes are necessary. Finally, GAPDH, TUB and β-Actin are unsuitable for normalization of qRT-PCR data under these given conditions in alfalfa. The relative expression level of MsWRKY33 was analyzed using selected reference genes. These results provide an experimental guideline for future research on gene expression in alfalfa using qRT-PCR.