This paper provides an introduction to the second stage ('part B') of an international exercise to determine the accuracy of current theories for predicting failure in composite laminates. It contains a detailed description of the process adopted for carrying out 'part B', so that the composites community can draw appropriate conclusions on the overall probity (and hence reliability) of this benchmarking task. Furthermore, in providing a clear description of the test cases and associated experimental data adopted for the exercise, the aim is for the information to be used by others as a standard benchmarking method for new failure prediction theories. (C) QinetiQ Ltd 2002.