For a [0/90] s cross-ply laminate subject to uniaxial loading, this paper presents the results of comparisons between the model predictions of stress/strain behaviour, carried out in Part A of the Failure Exercise, and the corresponding experimental data provided after the predictions had been made. In addition, because a new model designed for general symmetric laminates has since been developed, comparisons are shown between the stress/strain predictions of the new model applied to quasi-isotropic and [± θ] s laminates (with θ=45 and 55°), and the corresponding experimental data provided to participants in the failure exercise. Very good correlations between the model and experimental stress/strain data have been achieved for both the cross-ply and quasi-isotropic laminates (subject to both uniaxial and biaxial loading), principally because ply cracking in a single orientation is expected to occur for a significant part of the stress/strain curves. The correlation for [± θ] s laminates is not as good because the model is not able to account for ply crack formation in both the + θ and − θ plies of the laminate. In addition, the model has not yet been modified to deal with mixed-mode ply crack growth, or with non-linear matrix behaviour which could lead to significant deviations between model predictions and experimental results at large strains. It is emphasised that the models used are not able by themselves to predict strength. Additional damage modes must be taken into account (e.g. fibre fracture) before predictions of strength can be made that are based on the modelling of physical micro-mechanisms associated with laminate failure. A rudimentary physically based approach to the prediction of laminate failure is made by considering fibre strain within the laminate in relation to the average failure strain of a single fibre. The ply cracking models that have been assessed in the paper, for cases where they are expected to be valid, have been shown to lead to good predictions of laminate stress/strain behaviour that is consistent with data provided for the Failure Exercise. The reliability of the models is such that they have good potential for being used as the basis of new design methodology to deal with initial ply crack formation in laminates, and the effects of ply cracking on non-linear stress/strain behaviour for parts of stress/strain curves where ply cracking occurs in a single orientation.