Nitrate pollution is a global environmental issue. Forests play an important role in altering hydrological processes and purifying water pollutants in rainfall and runoff. The quantitative identification of nitrate concentration and sources in surface runoff is of great significance for watershed management and water environment improvement. In this study, water quality of surface runoff was monitored in three typical forest types in subtropical eastern China: Phyllostachys pubescens, Cunninghamia lanceolate, and Cyclobalanopsis glauca. Combined with hydrochemical analysis, we adopted the dual isotope approach (δ15N-NO3– and δ18O-NO3–) and Bayesian model (SIAR) to identify nitrate sources in forests that are subject to low anthropogenic disturbance. Results showed that the temporal variability of NO3-N concentrations was similar for all forest types, with higher values in periods of low rainfall and lower values in heavy rainfall periods. The NO3–-N concentration in runoff was much higher in C. glauca forests relative to P. pubescens and C. lanceolata. Both the Cl− concentrations and NO3–/Cl− molar ratio suggested the fertilizer inputs was the dominant source of nitrate in surface runoff. In agreement, δ15N-NO3– and δ18O-NO3– values inferred atmospheric deposition and chemical fertilizers to be the main sources of nitrate in all forest types. The Bayesian model outputs demonstrated that atmospheric deposition was the main source in the runoff in P. pubescens and C. lanceolate forests, contributing 28.83% and 35.08% to the total nitrate, respectively. In contrast, chemical fertilizers were identified as the main source in C. glauca forests, with NH4+ fertilizers and NO3– fertilizers accounting for 27.07% and 24.83%, respectively. Both chemical and isotopic analysis indicated that nitrification had little contribution to nitrate concentrations in runoff. Our results suggest that, even in forests with low anthropogenic disturbance, the application of fertilizer in surrounding agricultural regions should be effectively managed to minimize watershed nitrogen contamination.