Focusing on the various severity of the victims' injuries appeared in different disaster areas, using the diverse final due date or sufferable duration that the casualties can stick to be rescued, we characterize different disaster area impacts and discuss the cross-regional emergency vehicle routing problem with time window. Firstly, a vehicle routing optimization model with minimizing the traveling time is developed, and a modified ant colony algorithm is designed to solve this model. Then, the impacts of differentiated disaster on the cross-regional emergency rescue routing optimization are discussed. Finally, taking decision-makers' risk attitude as an example, we perform the simulation analysis and demonstrate the effects of key parameters. The results provide some suggestions for establishing an effective emergency management system.