The objective of this study is to validate a recently developed agent-based pedestrian simulation model, using data collected at the pedestrian walkway of Brooklyn Bridge. Video data were collected at the walkway and the trajectories of 294 pedestrians were extracted using computer vision. A genetic algorithm was applied to identify the optimum model parameters that minimize the error between the simulated and the actual trajectories of the calibration dataset. The simulation model was then applied to reproduce the trajectories of 214 pedestrians, considered for validation. The validation results showed that the model was capable of producing pedestrian trajectories with high accuracy, as the average location error between actual and simulated trajectories was for 0.32 m, while the average speed error was 0.06 m/s. Macroscopic results of the model were assessed by comparing the density–speed relationship in both actual data and the simulation. Finally, the accuracy of the model in reproducing the actual behavior of pedestrians during different interactions was evaluated. Results showed that the model was capable of handling these interactions with high accuracy, ranged between 79% and 100%.