Abstract: To further enhance the impact resistance of engineered cementitious composites (ECC), this study investigates the impact resistance of ECC reinforced with non-metallic fibers and steel fibers (SF) of varying elastic moduli and lengths through drop-weight impact tests. The results indicate that the number of impact failures of ECC increases with increasing fiber modulus and fiber length. Under the same volume fraction of non-metallic fibers, the number of impact failures for ECC reinforced with 12 mm polyethylene (PE) fibers is 2. 92 times and 1. 43 times that of ECC reinforced with 12 mm glass fibers (GF) and 36 mm glass fibers, respectively. The failure mode of ECC without fiber reinforcement is fragmentation into several pieces, whereas hybrid fiber-reinforced ECC only experiences local punching shear failure at the impact site, maintaining high integrity of the specimen. An impact damage evolution equation and a life prediction model based on the two-parameter weibull distribution were established. Comparing the prediction accuracy of machine learning methods such as k-nearest neighbors (KNN), particle swarm optimization random forest (PSO-RF), and particle swarm optimization k-nearest neighbors (PSO-KNN) with traditional fitting models, the PSO-KNN model demonstrates the best fit. The findings of this study provide a reference for selecting fibers in engineering applications and utilizing machine learning methods to predict the impact resistance of ECC.