An important failure mode of composite laminates is transverse cracking in the plies as a result of an applied load perpendicular to the fibre direction. This is a matrix-dominated mode. Transverse cracking in a composite can take place at loads below the failure load of the fibres. The damage sometimes leads to catastrophic failure and to property degradation of the laminates. A better understanding of this failure mode is particularly desirable. There has been a fair amount of work [1-7] on the mechanical behaviour of transverse cracking. It has been shown that the transverse crack process is controlled by the fracture toughness, defined by the critical strain energy release rate Gc, associated with this type of failure. The measurement of G e provides basic information about the transverse crack resistance of the laminates. In addition, it provides the materials scientist with an index for selecting resin matrix materials for composites to better resist transverse cracking. There are numerous test methods for measuring the fracture toughness of composite laminates. Most of them are based on tests originally used for homogeneous materials. The application of these techniques to angle-ply and cross-ply composites has not been all that satisfactory because of the complex damage phenomena involved. These composites do not usually fail with a single crack as is usually the case with homogeneous materials. Different methods have been developed to test laminates for transverse cracking parallel to the fibres. For example, Sanford and Stonesifer [8] used double edge notched (DEN) and single edge notched (SEN) specimens for evaluating the fracture toughness of (0 °) fibreglass laminates. Parvizi et al. [5] also used DEN and notched three point bend specimens for fracture toughness measurement on (0 °) fibreglass laminates. Slepetz and Carlson [9] employed compact tension specimens to test (0 °) S-glass/epoxy and (0 °) graphite/epoxy composites. The compliance calibration technique has been most conveniently employed for measuring the Gc of composites. The specimen compliance, C, against crack length, a, relationship is first established. In order to obtain an accurate C-a curve, a large enough population of specimens must be tested. In this respect, a test method which yields a linear C-a relationship is certainly more convenient to use, for example, the double torsion (DT) test [10, 11 ]. In addition to the advantage of rapid calibration of the C-a relationship, the test requires very simple specimen geometries which greatly minimize the machining operation involved. In this study, the double torsion test was investigated for its applicability to characterize transverse cracking in unidirectional fibre composites. This method has been successful for measuring fracture toughness of brittle homogeneous materials such as glass [10], ceramics [11] and glassy polymers [12-14]. When the technique was applied here to laminates, the load-displacement curves of the specimens were found to be different from those of the homogeneous materials. A correct interpretation of the results, however, led to the determination of meaningful fracture toughness related to transverse cracking. The original double torsion test technique was suggested by Outwater and developed by Kies and Clark [10]. The specimen is a rectangular plate supported along its two longer edges. In our case, the fibre direction was along the longer axis of the specimen. The specimen can be supported by either two rollers (Fig. 1) [11-14] or four steel spheres [15, 16]. In this study, we have experimented with test fixtures built with both of these supporting conditions. The load was applied to the specimen through two steel spheres as shown in Fig. 1.