NeuroMiner has been continuously developed by Nikolaos Koutsouleris since 2009 (with support from the European Commission for the PRONIA project since 2013) to provide clinical researchers with cutting-edge machine learning methods for the analysis of heterogeneous data domains, such as clinical and neurocognitive information, structural and functional neuroimaging data, and genetic markers. The program can be considered as an interface to a large variety of unsupervised and supervised pattern recognition algorithms that have been developed in the machine learning field. Furthermore, the program implements different strategies for preprocessing, filtering and fusing heterogeneous data, training ensembles of predictors, and visualizing and testing the significance of the computed predictive patterns. The current release of NeuroMiner has been tested in the Section of Neurodiagnostic Applications on a variety of datasets from healthy controls and patients with psychiatric disorders and was designed specifically to create robust models with a high probability of generalization to new datasets.