The normalization of data by choosing suitable reference genes is fundamental for obtaining accurate and reliable results in quantitative real-time polymerase chain reaction (qPCR) analyses. In this study, the expression stability of 12 candidate reference genes of Pinus massoniana under different abiotic stresses was evaluated using four statistical algorithms: geNorm, NormFinder, BestKeeper, and RefFinder. The results indicate that the following genes could be used as reference genes under different treatments: Actin 2 (ACT2) and F-box family gene (F-box) for salinity treatment, cyclophilin (CYP) and alpha-tubulin (TUA) for ABA treatment, actin 7 (ACT7) and CYP for drought treatment, actin 1 (ACT1) and ACT7 for cold treatment, ACT1 and CYP for heat treatment, and TUA and ACT2 for the “Total” group. To validate the suitability of the selected reference genes in this study, the Short-Root protein (SHR), Alpha-pinene synthase (APS), and Pyrabactin resistance-like protein (PYL) gene expression patterns were analyzed. The expression patterns had significant biases when the most unstable reference genes were used for normalization, compared with when the optimum reference gene or gene combinations were used for normalization. These results will be beneficial for further studies on gene transcription in early-stage, unlignified seedlings of P. massoniana.