This study, using artificial neural networks, support vector machines as tools of machine learning derived from artificial intelligence (AI), multivariate discriminant analysis (MDA) and logistic regression (LR), assesses the role of financial ratios, firms' characteristics, and macroeconomic indicators in predicting financial distress among Egyptian small and medium-sized firms (SMEs). Our empirical findings reveal that combining financial variables with the variables of firms' characteristics (age and industry) increases the accuracy of predicting financial distress among firms of this kind. However, the inclusion of macroeconomic information has no impact on the predictive accuracy of neural networks. Moreover, in a comparison we also assess the predictive accuracy of multilayer perceptrons (MLPs) to support vector machines (SVM), and other traditional statistical techniques. According to the benchmarking results of the MDA, LR, SVM and MLP models, the neural network model (MLPs) outperforms MDA, LR and SVM as regards the predictive accuracy of the out-of-sample set.