We describe various industrial and business applications of a common approach for learning the probability distributions and the structure of Bayesian networks. After describing the Bayesian learning theory, we explain how the parameters learned with this method can be used for prediction tasks in various application domains. In the first domain we learn the structure and the probability distribution of a dynamic Bayesian network to diagnose faults in a generic electrical power distribution network. In the second domain, we learn the probability distributions of a Bayesian network in a multi-agent system for the profiling of users in the procurement and contracting for the USA defense acquisition community. In the third application we describe a help-desk on-line information system for managing a hotel chain using Bayesian reasoning to assign the best technician to answer client requests. With these applications we show how a theoretical Bayesian framework can be integrated into on-line information systems and scaled-up to solve real world problems.