Electric impact wrench is an important tool used in manufacturing and maintenance services. It has complex mechanism and its operation involves dynamic events occurring at vastly different time scales, which poses challenges for efficient and accurate modeling to facilitate design optimization and control. This investigation establishes a first principle-based, system-level model of a representative impact wrench. The model explicitly incorporates the dynamic flexibility of gear transmission, spindle shaft, and impacting components into the kinematic relations that connect them together. The nonlinear impact and contact events, coupled with the rotational and translational motions of all components, are explicitly analyzed, and systematic parametric identification is performed based on a multi-objective optimization (MOO) approach. The model prediction is correlated with experimental studies.