ncPred is an algorithm for the inference of ncRNA-disease association based on recommendation technique. We represent our knowledge through a tripartite network, whose nodes are ncRNAs, targets, or diseases. Interactions in such a network associate each ncRNA with a disease through its targets. Our algorithm, starting from such a network, computes weights between each ncRNA-disease pair using a multi-level resource transfer technique that at each step takes into account the resource transferred in the previous one.