[Objective] The effectiveness of ecological buffer zones (EBZs) in mitigating non-point source pollution is influenced by multiple internal and external factors, whose relative significance remains incompletely understood. [Methods] Using an orthogonal experimental design, this study systematically evaluated the purification capacity of EBZs for nitrogen (TN) and phosphorus (TP) pollutants in both surface runoff and subsurface flow. Two internal factors (vegetation configuration and slope gradient) and two external factors (influent pollutant concentration and hydraulic loading rate) were investigated. Range analysis was applied to quantify the relative contributions of these factors to pollutant removal efficiency. [Results] The findings reveal that EBZs exhibited significantly higher removal efficiencies for TN and TP in subsurface flow (21.31% and 29.80%, respectively) than in surface runoff (16.91% and 23.04%). Composite vegetation buffers (particularly shrub+herb configurations) outperformed monoculture herbaceous strips, achieving average removal rates of 20.16% (TN) and 27.24% (TP) in runoff, and 24.54% (TN) and 33.89% (TP) in subsurface flow. Factor influence hierarchy was: vegetation configuration (most dominant) > influent concentration > slope gradient > flow rate (least influential). Pollutant removal efficiency decreased with increasing slope gradient, flow rate, and influent concentration across all vegetation types. [Conclusion] Through orthogonal testing and range analysis, this study quantitatively ranks the effects of key factors on EBZ performance. The results provide actionable technical guidance for optimizing EBZs designs to control agricultural non-point source pollution.