For the purpose of studying wildfire spread in living vegetation such as chaparral in California, a thermal particle image velocity (TPIV) algorithm for nonintrusively measuring flame gas velocities through thermal infrared (IR) imagery was developed. By tracing thermal particles in successive digital IR images, the TPIV algorithm can estimate the velocity field in a fire-plume flow. In this method, the thermal particle is assumed to be a virtual particle that corresponds to the "hot" pixel resolved in a digital image. Its signal is a temperature value measured by a thermal IR camera. Fire-plume experiments using alcohol and chamise ( Adenostoma fasciculatum ) as fuels and a computational simulation have been completed. A thermal IR camera was used to capture successive fire temperature images. Using these images, velocity fields were estimated by the TPIV algorithm. For numerical simulation, the computed detailed instantaneous velocity field was used to validate the field estimated by the TPIV algorithm. Results illustrate the applicability and performance of the TPIV algorithm in fire research.