The Hazard Prediction and Assessment Capability (HPAC) model system, which contains the SCIPUFF dispersion model, is widely used by the U.S. Department of Defense and other agencies. Version 5.0 SP1 (January 2008) of HPAC has been significantly updated to improve urban wind speed profile estimates and to add a model to calculate detailed flow and dispersion around individual buildings (MicroSwift/Spray, or MSS). Evaluations with urban tracer data sets are underway, and the most recent evaluations presented in this paper use data from the Madison Square Garden 2005 (MSG05) field experiment, which included detailed supporting meteorology and six different perfluorocarbon tracers (PFTs) released during four periods on two days in March, 2005. The MSG domain contains several very tall buildings with heights greater than 150 m, and is marked by deep street canyons. HPAC/SCIPUFF is run using three different urban dispersion module options (Urban Dispersion Model (UDM), Urban Canopy (UC), and MSS) and three different input meteorology options (observed average winds, basic default airport, and mesoscale meteorological model). For comparison purposes, a simple Gaussian-format urban dispersion model has also been run for the same MSG05 data base. The evaluations focus on hourly-averaged concentrations paired in time and space. It is shown that the urban HPAC/SCIPUFF options produce reasonable performance, although similar performance is also found for the simple urban dispersion model.