REliable PIcking by Consensus (REPIC) is an ensemble learning approach to cryogenic-electron microscopy (cryo-EM) particle picking. It identifies particles common to multiple picked particle sets (i.e., consensus particles) using graph theory and integer linear programming (ILP). Picked particle sets may be found by a human specialist (manual), template matching, mathematical function (e.g., RELION's Laplacian-of-Gaussian auto-picking), or machine-learning method. REPIC expects particle sets to be in BOX file format (*.box) where each particle has coordinates, a detection box size (in pixels), and (optional) a score [0-1].