RISO is an implementation of heterogeneous, distributed belief networks in Java. A belief network is a probability model defined on an acyclic directed graph; distributed means nodes can be on different hosts, and heterogeneous means allowing different conditional distributions. The calculations involved are multidimensional integrations; exact results are known for a catalog of special cases. If a partial result cannot be calculated as a special case from the catalog, RISO computes an approximate result by numerical integration. Partial results are passed from one node in the graph to another as messages; if nodes live on different hosts, the belief network is said to be distributed. Messages are passed via RMI. Many example belief networks and lengthy documents are included in the RISO release bundle.