Under some circumstances, it may be desirable to provide all or part of a building with collective- protection against harmful chemical or biological (CB) agents. Collective-protection, as opposed to individual protection, uses the building -- its architecture, ventilation system, and control components -- to safeguard the health of the building occupants in the event of an indoor or outdoor release of toxic agents. In this study, we investigate the movement of tracer gases within a six-story building. The building was retrofitted to provide collective-protection on the upper two floors. To achieve this protection, the upper floors were over-pressurized using outside air that had passed through military specification carbon canisters and high-efficiency particulate air (HEPA) filters. The four lower floors were outside the collective-protection area and had a ventilation system that was retrofitted to provide response modes in the event of a CB release. These response modes (e.g. building flush and shelter in place) were designed to reduce the exposure of occupants on the lower floors without compromising the collective-protection zones. Over the course of four weeks, 16 tracer gas experiments were conducted to evaluate the collective- protection system (CPS) of the building's upper two floors and the ventilation response modes of the lower floors. Tracer gas concentrations were measured at a rate of 50 Hz in up to 30 locations in each experiment, which provided data with very high spatial and temporal resolution. Differential pressure and temperature measurements were also made throughout the building. Experiments showed that the CPS maintained a positive pressure differential between the upper two floors and the lower floors with various meteorological conditions and within specified settings of the HVAC fans serving the lower floors. However, the tracer experiments did show that a CB agent could enter the first zone of the decontamination areas on each CPS floor. Tracer gas analysis also showed that the shelter in place HVAC mode provided protection of lower floor occupants from an outdoor release by significantly lowering the air exchange rates on those floors. It was also determined that the efficacy of a flush mode triggered by an agent sensor depends greatly on the location of the sensor.