This paper presents a Bayesian modeling technique, called empirical Bayesian finite response (EBFIR) modeling, that helps deal with the collinearity problem usually encountered in FIR models, and helps improve the estimation accuracy of their coefficients. The developed technique iteratively solves for the prior density used in estimation and the FIR coefficients. The advantages of the developed EBFIR modeling technique are also illustrated though a simulated example.