Groups of size three completed an estimation task by interacting either face-to-face or via a videoconferencing system. Results indicate that there are no significant differences between the two interaction media on the following outcome dimensions: accuracy; over-confidence; size of credible intervals; or improvement over average initial individual estimates. Nevertheless, significant differences were found in the confidence attached by groups to their decisions. Compared to videoconferencing groups, face-to-face groups showed greater increases in confidence in their decisions, especially if they were instructed to discuss their beliefs and assumptions underlying their estimates and not the estimates themselves. However, this confidence was not as appropriate as that in the videoconferencing groups. Further, videoconferencing groups reported modifying more of their beliefs during discussion. Implications for the design and application of advanced systems for decision-making support and research are discussed.