The geotechnical conditions at Mexico City suggest strongly the idea of decoupling the vertical resonance of the thin, very soft, clay layer at the surface, from the two-dimensional (2-D) effects of a much stiffer underlying sedimentary valley. The purpose of this paper is to explore whether numerical modeling supports this idea. This problem has also a bearing on the contradiction between the extent of local models usually assumed by seismologists and those considered by geotechnical engineers. We present results of 1-D models and of two different approaches to a 2-D modeling in an effort to explore the extent of local geology that must be considered in evaluating site effects at Mexico City. Our results indicate that there is a significant degree of coupling between the surficial clay layer and the underlying sediments, despite the extremely large impedance contrast. As a consequence, a full 1-D analysis gives better results than a 2-D “decoupled” one. It is also shown that deeper layers affect significantly ground motion at the surface. Thus, a simplified 1-D model that would consider only the uppermost very soft layers would be wrong in its prediction of surface motion by a factor about 3 both in frequency and time domains.