The presence of open holes in composite structures increases stress concentration and reduces load carrying capacity. By analyzing data of compression strength experiment, a prediction model based on generalized regression neural network of compression strength of a laminate with open hole is proposed. This model uses the diameter of the open hole, as well as the length and width of the laminate as input parameters, and the compression strength as the output parameter. The Gauss function is used as the transfer function. The model uses part of the experimental data to train for finding the optimal smooth factor. The optimal smooth factor was calculated using the S-fold cross-validation method. The results indicate that the model, based on a generalized regression neural network, has good generalization ability for the test data. It can be used to estimate the residual strength of composite laminates in open hole compression.