Quantitative real-time reverse transcription PCR (qRT-PCR), a sensitive technique for quantifying gene expression, depends on the stability of the reference gene(s) used for data normalization. Only a few studies on the reference genes have been done with peanut to date. In the present study, 14 potential reference genes in peanut were evaluated for their expression stability using the geNorm and NormFinder statistical algorithms. Expression stability was assessed by qRT-PCR across 32 biological samples, including various tissue types, seed developmental stages, salt and cold treatments. The results showed that the best-ranked references genes differed across the samples. UKN1, UKN2, TUA5 and ACT11 were the most stable across all the tested samples. A combination of ACT11, TUA5, UKN2, PEPKR1 and TIP41 would be appropriate as a reference panel for normalizing gene expression data across the various tissues tested, whereas the combination of TUA5 and UKN1 was the most suitable for seed developmental stages. TUA5 and EF1b exhibited the most stable expression under cold treatment. For salt-treated leaves, TUA5 and UKN2 were the most stably expressed and HDC and UKN1 for salt-treated roots. The relative gene expression level of peanut Cys(2)/His(2)-type zinc finger protein gene AhZFP1 was analyzed in order to validate the reference genes selected for this study. These results provide guidelines for the selection of reference genes under different experimental conditions and also a foundation for more accurate and widespread use of qRT-PCR in peanut gene analysis.