Response surface methods or metamodels are commonly used to approximate large engineering systems. This paper presents a new metric for evaluating a response surface method or a metamodeling technique. Five response surface methods are studied: Stepwise Regression, Moving Least Square, Kriging, Multiquadratic, and Adaptive and Interactive Modeling System. A real world frontal impact design problem is used as an example, which is a complex, highly nonlinear, transient, dynamic, large deformation finite element model. The optimal Latin Hypercube Sampling method is used to distribute the sampling points uniformly over the entire design space. The Root Mean Square Error is used as the error indicator to study the accuracy and convergence rate of the metamodels for this vehicle impact analysis. A hybrid approach/strategy for selecting the best metamodels of impact responses is proposed.