The need for efficient tools to study the transport and dispersion of chemical, biological, or nuclear (C/B/N) agents has been the center of attention for more than a decade. The increasing computational power combined with the improvement of algorithms has made CFD models attractive tools to study atmospheric releases at urban scales. However, these tools have not reached the desired rapidness yet. If the time frame provided by the National Research Council (NRC) is considered (immediate first response, 0 to 2 hours; early response, 2 to 12 hours; and sustained response support, greater than 12 hours), CFD tools can be expected to perform in the two upper brackets of this classification: early and sustained response. There have been attempts to make CFD usable in the immediate response time, but these approaches usually rely on the pre-calculations of the situations with interpolated and/or the simplified geometrical details of the modeled area. One common simplification is to consider the terrain as a flat surface ignoring all elevation differences on the landscape. Although this approach has been used in the past, the present work suggests that such approach could provide the end user with wrong answers for the calculated concentration levels. To illustrate this important aspect of CFD modeling, we will present the building reconstruction of the main campus of George Mason University in Fairfax, VA. We will analyze the release of a passive agent in the campus, and will study the transport and dispersion of the scalar. Two different cases will be examined in this paper: initially the campus terrain will be considered as a flat surface, and then, the realistic terrain description will be integrated into the computational model. Results show dramatic differences in modeled concentration levels in selected locations.