Bench-scale tests measuring the thermal protective performance of textile materials do not capture the effect of thermal shrinkage, primarily because of the planar geometry of the test device. The performance of single-layer fabrics commonly used in protective garments is compared here following several protocols, including the use of a new cylindrical device as well as standard and modified ASTM, CGSB and ISO procedures, with and without a 6.35 mm air gap between the fabric and the sensor. Both the time to reach the second degree burn criterion and the time for the sensor to register a 24degreesC temperature rise were measured. Fabrics that shrink had reduced thermal protection when measured with the cylindrical device, compared with other tests. Two-way analysis of variance indicated that, although the dependent measures differ significantly among fabrics, the nature and extent of those differences depend on the test used. One-way analyses of variance indicate that each method differentiates among the fabrics. However, most tests in which the fabrics are in contact with the sensor rank heaviest fabrics as the most protective. Among the tests incorporating a space between the fabric and sensor, those using the cylindrical device differentiate best between lighter fabrics that shrink and those that do not. Regression analyses of data from bench scale tests with data from instrumented mannequin tests confirm the superiority of the cylindrical device in capturing the effects of both thermal shrinkage and fabric integrity. Copyright (C) 2002 John Wiley Sons, Ltd.