This study examines the implications of the short-run specification of the standard, static translog cost function along with the possible implications of non-stationarity by estimating a dynamic translog cost specification complete with dynamic share equations for the U.S. using an empirical approach developed by Urga and Walters (2003). We compare the results of the static, long-run model with those of a dynamic, short-run error-correction model in terms of 1) significance of the parameter estimates, and 2) consistency with economic theory.