BayesRCO is a software for complex trait prediction with Bayesian hierarchical models using genome-wide genetic variation grouped into potentially overlapping annotation categories based on prior biological information (e.g., functional annotations, candidate gene lists, known causal variants). BayesRCO includes implementations for three state-of-the-art Bayesian hierarchical models: - BayesCpi: a two-class model, corresponding to null and non-null effects for genetic variants - BayesR: a four-class model, corresponding to null, small, medium, and large effects for genetic variants - BayesRC: a BayesR model incorporating disjoint prior categories for genetic variants. In addition, BayesRCO includes two novel extensions of BayesRC to incorporate potentially overlapping prior categories for genetic variants: - BayesRC+: a BayesR model where multi-categories are assumed to cumulatively impact variant estimates - BayesRCpi: a BayesR model where the catégorization of multi-annotated variants is stochastically modeled.