BackgroundSoybean pathogens and pests reduce grain production worldwide. Biotic interaction cause extensive changes in plant gene expression profile and the data produced by functional genomics studies need validation, usually done by quantitative PCR. Nevertheless, this technique relies on accurate normalization which, in turn, depends upon the proper selection of stable reference genes for each experimental condition. To date, only a few studies were performed to validate reference genes in soybean subjected to biotic stress. Here, we report reference genes validation in soybean during root-knot nematode (Meloidogyne incognita) parasitism and velvetbean caterpillar (Anticarsia gemmatalis) attack.FindingsThe expression stability of nine classical reference genes (GmCYP2, GmELF1A, GmELF1B, GmACT11, GmTUB, GmTUA5, GmG6PD, GmUBC2 and GmUBC4) was evaluated using twenty-four experimental samples including different organs, developmental stages, roots infected with M. incognita and leaves attacked by A. gemmatalis. Two different algorithms (geNorm and NormFinder) were used to determine expression stability. GmCYP2 and GmUBC4 are the most stable in different organs. Considering the developmental stages, GmELF1A and GmELF1B genes are the most stable. For spatial and temporal gene expression studies, normalization may be performed using GmUBC4, GmUBC2, GmCYP2 and GmACT11 as reference genes. Our data indicate that both GmELF1A and GmTUA5 are the most stable reference genes for data normalization obtained from soybean roots infected with M. incognita, and GmCYP2 and GmELF1A are the most stable in soybean leaves infested with A. gemmatalis.ConclusionsFuture expression studies using nematode infection and caterpilar infestation in soybean plant may utilize the reference gene sets reported here.