Quantifying the ventilation available to an enclosure fire is an important step to understanding fire behavior. Accurate measurements of mass flow rate across an enclosure opening require a complete mapping of the velocity and density fields due to the three dimensional nature of vent flows. Conventional flow measurement methods in fire research consist of vertical arrays of thermocouples and differential pressure probes at the vent plane which are physically obtrusive and offer limited spatial sampling. A reduced-scale analog of a full-scale fire experiment was studied for the purpose of studying the potential use of Stereoscopic Particle Image Velocimetry, a laser based nonintrusive imaging technique, to measure fire induced flows through vents. The experiment was isothermal and modeled the convective transport by using a helium plume as the buoyant source. Stereoscopic PIV measurements were successfully demonstrated for a large-scale flow field with planar image regions of 0.71 m x 0.62 m (l x h). Measurements of the complete velocity vector, v x ,v y ,v z , were performed and a full mapping of the velocity field in the region of the doorway was achieved. The vector field data displays the three dimensional structure of the flow through the doorway, revealing regions where the velocity component normal to the doorway plane may not completely dominate the velocity vector. A comparison of mass flow rate computations using the velocity component normal to the opening, v x , and the velocity magnitude, demonstrated that mass flow rate will be over predicted by as much as 25 % if computed using the velocity magnitude. Velocity magnitude is representative of bi-directional probe measurements and the over prediction is consistent with the use of doorway flow orifice coefficients to correct mass flow rates computed from bi-directional probe data. The intermediate scale of the flow field was sufficient to test the performance of a current PIV system and to identify the requirements for conducting successful PIV measurements in a full-scale fire experiment.iii ACKNOWLEDGEMENTS