This paper explores various inference techniques for an intelligent decision support system based on influence diagrams. Rule—based expert systems for decision support have been successful for well-structured, well understood decision situations of a taxonomic classification type. As uncertainty is prevalent, information costly and payoff relevant, and the preferred solution depends on the specific beliefs and preferences of an individual or group decision maker the resolution methods of decision theory embodied in first—order predicate logic forms a natural basis for computerised intelligent decision support. Based on a unified characterisation of knowledge inference procedures for logical probabilistic and decision theoretic reasoning are described in detail.