Many railway tank-cars carrying hazardous materials are thermally protected from fire impingement by thermal insulation and a steel jacket applied to the outside of the tank-car shell. Over time, it is possible that the thermal insulation will sag, rip, degrade, or be crushed under the steel jacket. A thermographic technique to determine whether or not a tank has insulation deficiencies has been developed, but it is necessary to determine which thermal deficiencies do not affect a tank's survivability in a fire and which thermal deficiencies must be repaired. In order to develop a guideline in assessing thermal defects, a thermal model and experimental data would be beneficial. A series of fire tests were performed on a quarter-section tank-car mock-up to assist in developing a guideline and to provide validation data for a thermal model. Twelve fire tests, with constant, credible, simulated pool fire conditions, were performed on the tank-car mock-up with various insulation deficiencies. An infrared thermal imaging camera was used to measure the tank wall temperature. The thermal images were useful in determining the temperature profiles across the defects at different times and the transient temperature behaviour at different locations. It was seen that the properly installed thermal protection system significantly reduced the heat transfer from the fire to the tank wall. It was also seen that the steel jacket alone (i.e. 100% defect) acted as a radiation shield and provided a significant level of protection. With small defects, it was observed that the surrounding protected material provided a cooling effect by thermal conduction. A square defect greater than about 40 cm on each side should be considered significant, because unlike smaller defects, there is little benefit from the surrounding material as far as the peak defect temperature is concerned. (C) 2003 Elsevier Ltd. All rights reserved.