This study presents a web-based group decision support system that considers the decision making process as a whole, including idea generation and sharing, planning, discussion, decision inference, and evaluation. The decision support infrastructure is based on agent and object-oriented technology. Two major functionalities are identified and developed—(i) an information classification and retrieval module that supports the collection of opinions and ideas sharing from an organization. Internal and external information is managed and can be retrieved efficiently by using structured searching functions; and (ii) a decision inference module in which decisions can be made either by a rule-based reasoning model for a certain situation or by a Bayesian network-based reasoning model for a probabilistic situation. Although these modules can also perform designed functions independently, they are indeed highly interrelated in a seamless way. In particular, the information classification and retrieval module collects information and guides decision-making at a meta level. From a knowledge management perspective, it structures the valuable information collected for effective reuse. Then, the users can use the decision inference module to derive plausible alternatives while referring to collected and classified information. The system has been implemented in a leading telecommunication company in Taiwan and run within an Internet-based environment, accessible by authorized decision makers without geographical constraints. This work details the development of this system as well.