Abstract: The partial nitrification/anammox (PN/A) process has attracted much attention in the field of wastewater nitrogen removal due to its advantages such as low energy consumption and low carbon source demand. However, the mechanism of the influence of influent chemical oxygen demand (COD) on its stability is still unclear. In this study, an 8 L sequencing batch reactor was used to establish a one-dimensional multi-population biofilm mathematical model to investigate the operating characteristics of the PN/A process under different COD loads (0-200 mg/L) and dissolved oxygen (DO) concentrations (0.1-5.0 mg/L). The model set a radial gradient inside the particles and fixed the influent ammonium concentration at 100 mg/L. The Monte Carlo method was used for parameter calibration and uncertainty analysis. The results showed that: (1) The influent COD concentration significantly affected the spatial distribution of functional bacteria. When COD was low (<50 mg/L), ammonia oxidizing bacteria were enriched in the outer layer of the particles, while high COD (>150 mg/L) led to the dominance of heterotrophic bacteria. (2) Moderately increasing the influent carbon-nitrogen ratio helped the system adapt to the DO concentration range and improve the flexibility and stability of operation. By strengthening the denitrification pathway, the contribution rate of nitrogen removal was increased by about 20%, and the total nitrogen removal efficiency of the system was stably maintained at more than 85%. (3) By improving the stoichiometric matrix, accurate tracing of the nitrogen oxide conversion pathway was achieved, confirming that the anammox process was still the core nitrogen removal pathway. This study elucidates the synergistic regulatory mechanism of influent COD and DO concentrations for achieving efficient operation, providing critical theoretical guidance for optimizing PN/A process parameters.