Decision making in individuals is subject to pervasive and predictable errors (as defined by Expected Utility Theory, EUT). The most sophisticated exposition of these effects is Kahneman and Tversky's Prospect Theory (PT). Several theorists have noted that PT might also apply to the decision-making processes of small groups, but this appears never to have been investigated before. This study tested 4 of the major PT biases (the certainty, reflection, isolation, and shift of reference effects) in 32 pre-existing committees (3–7 members) of Southern Baptist churches, responsible for important church decisions. Committees were presented with 6 decision-making tasks (e.g., fund raising strategy, new pastor selection) to test whether the PT effects operate in small groups. In addition, the Group Environment Scale was administered to assess the effects of group cohesiveness, leadership, task orientation, and organization. Group discussions were analyzed for themes relevant to decision-making. In general, group decisions were influenced by the certainty and reflection effects, although not by isolation or shift of reference, perhaps because the tasks designed to test them did not make the effects salient. Other strong factors determining decisions were identified. A model of decision making in such groups was proposed in which a hierarchy of heuristics is considered; in this case: (a) avoid a morally repugnant choice, e.g., passivity, (b) avoid confusion or uncertainty, (c) avoid or minimize loss, and (d) maximize gain. If a higher-level heuristic was invoked to guide the decision, lower-level ones were largely irrelevant. It was proposed that this hierarchy could explain not only decisions that run counter to the predictions of EUT, but also the minority of groups and individuals whose decisions cannot be explained by either EUT or PT, which have largely been ignored in the literature. Group dynamics variables had virtually no effect on either the decisions or participants' ratings of their satisfaction with the outcome, except for a minimal positive effect of leader control. Analysis of group discussions confirmed that many decisions were evaluated on a risk vs. gain dimension, but there was little evidence of other psychological processes proposed by PT to explain the biased decisions.