In the event of a toxic chemical release to the atmosphere, shelter-in-place (SIP) is an emergency response option available to protect public health. This paper is the last in a three-part series that examines the effectiveness of SIP at reducing adverse health effects in communities. We model a hypothetical chemical release in an urban area, and consider SIP effectiveness in protecting occupants of commercial buildings. Building air infiltration rates are predicted from empirical data using an existing model. We consider the distribution of building air infiltration rates both with mechanical ventilation systems turned off and with the systems operating. We also consider the effects of chemical sorption to indoor surfaces and nonlinear chemical dose-response relationships. We find that commercial buildings provide effective shelter when ventilation systems are off, but that any delay in turning off ventilation systems can greatly reduce SIP effectiveness. Using a two-zone model, we find that there can be substantial benefit by taking shelter in the inner parts of a building that do not experience direct air exchange with the outdoors. Air infiltration rates vary substantially among buildings and this variation is important in quantifying effectiveness for emergency response. Community-wide health metrics, introduced in the previous papers in this series, can be applied in pre-event planning and to guide real-time emergency response.