Plant water source is a prerequisite for research and management of agriculture and ecology. Especially in Karst areas, due to the special geological conditions, plants are prone to generally suffer in severe water deficit. Understanding the plant water sources is therefore important for ecological restoration. In this study, the main dominant plants in the secondary forest of southwest karst regions, Kalopanax septemlobus (Thunb.) Koidz., Toona sinensis and Platycarya strobilacea Sieb.et Zucc. were selected. Isotopic samples of soil moisture at different soil depth, spring water, rain water and plants were collected. We analyzed the plants water sources by IsoSource and MixSIAR models, and the performance of the two models were compared. The results showed that the δ¹⁸O values of rainfall were positive during March to June, while these values were negative during June to September, 2017. Thus, the δ¹⁸O of rainfall exhibited the significantly temporal or seasonal variations. The δ¹⁸O values of soil moisture at different soil layers decreased with the increase of soil depth in spring, while this circumstance was contrary in summer. There were significant differences in calculating the proportion of plants water sources between IsoSource and MixSIAR models. The analysis of plant water sources based on the direct inference approach showed that the performance of MixSIAR model was better than that of IsoSource model. The performance of MixSIAR model (Root Mean Square Error (RMSE), 0.61 in spring and 0.59 in summer) outperformed the IsoSource model (RMSE, 0.84 in spring and 0.74 in summer) in estimating the plants water sources. The results of the study can provide a beneficial guide in model decision for the future researchers in plant water sources.