There is growing interest in data-driven analysis, multivariate statistics and predictive modeling for neuroimaging. Datasets are also constantly growing in sample size, resolution, and complexity. Nilearn is a Python package designed to address these new challenges in contemporary data analysis for imaging neuroscience. It provides state-of-the-art machine-learning methods for convenient pre-processing, analysis, and visualization of various types of neuroimaging results (i.e., experimental fMRI, VBM, and resting-state correlations). The tutorial will cover: (i) plotting and image manipulation with nilearn; (ii) decoding and predictive models; (iii) functional connectivity and resting-state data analysis.