<p>Urban runoff poses a significant management challenge due to its potential to damage infrastructure and communities. Accurate runoff modeling is essential for flood mitigation and urban planning. In this study, a Cellular Automata (CA) model was developed to simulate urban runoff. The model uses three transition rules to identify flow-receiving neighbors, distribute flow, and calculate cell flow, considering elevation, precipitation, buildings, and infiltration. Optimal parameters for pixel size, neighborhood configuration, and the flow-direction algorithm were determined. The model was applied to the urban area of Karaj City and validated against FLO-2D. Results show that rasterizing vector data using a 1-m cell size improved RMSE by about 40% compared with a 30-m cell size, while using eight instead of four neighbors improved RMSE by about 27%. Applying the minimization-of-differences algorithm reduced RMSE by about 10% compared with a steepest-direction approach. Moreover, primary streets were the most vulnerable, and impervious land uses generated higher runoff and larger inundated areas than green spaces. Robustness was evaluated through validation under an additional rainfall event, together with sensitivity and uncertainty analyses. The proposed CA model demonstrated good agreement with FLO-2D. Its simplicity and low data requirements make it a practical tool for urban runoff modeling.</p>