A new measure called minimum findable temperature difference (MFTD) is offered as a means of characterizing thermal imaging performance under scene clutter limited conditions. MFTD extends an older performance methodology called minimum detectable temperature difference (MDTD) which was offered in the 70's as a sensor performance characterization under noise limited conditions. MFTD departs from MDTD in that it addresses scene clutter plus noise as well as sensor-noise-only operating conditions. In addition, MFTD measures the in-FOV search performance of the thermal imager. This paper details the measurement procedures associated with MFTD using real target vehicles and equivalent bar targets. Methods of characterizing thermal target scenes in terms of vehicle signatures and scene clutter are offered that appear to correlate with measured probability of target finding. Methods used to train MFTD observers and procedures for processing displayed target embedded scenes are provided. Finally, applications of MFTD for modelling manned sensor performance in cluttered environments using TTPF model formulations are offered. A future application of MFTD with real targets to develop standard Army observer ROC curves is suggested as well.