Intelligent user interfaces, particularly in interactive group settings, can be based on system explanations that guide model building, application, and interpretation. Here we extend Silver's (1990,1991) conceptualization of decisional guidance and the theory of breakpoints in group interaction to operationalize feedback and feedforward for a complex multicriteria modeling system operating within a group decision support system context. We outline a design approach for providing decisional guidance in GDSS and then test the feasibility of the design in a preliminary laboratory experiment. Findings show how decisional guidance that provides system explanations at breakpoints in group interaction can improve MCDM GDSS usability. Our findings support Dhaliwal and Benbasat's (1996) conjecture that system explanations can improve decisional outcomes due to improvement in user understanding of decision models. Further research on intelligent agents, particularly in interactive group settings, can build on the concepts of decisional guidance outlined in this paper.