A person may reasonably accept some experimental report, hypothesis or theory because there is a consensus among an appropriate reference group of experts. It may be unreasonable, moreover, for a person to accept such statements when there is a consensus against such acceptance.1 A person may, however, conclude on the basis of careful study that the ex perts are in error. Having concluded thus, he may reasonably dissent from the experts, refusing to accept what they do, or accepting what they do not. For such a man, dissensus is reasonable and conformity counter productive. When is it reasonable for a person to conform to a consensus and when is it reasonable for him to dissent? We shall answer the question posed in terms of an intellectual concern of science and rational inquiry. Succinctly stated, the concern is to obtain truth and avoid error. We shall argue that consensus among a reference group of experts thus concerned is relevant only if agreement is not sought. If a consensus arises unsought in the search for truth and the avoidance of error, such consensus provides grounds which, though they may be overridden, suffice for concluding that conformity is reasonable and dissent is not. If, however, consensus is aimed at by the members of the reference group and arrived at by intent, it becomes conspiratorial and irrelevant to our intellectual concern. 1. Reasonable Acceptance and Expected Utility. To understand the role of consensus for reasonable acceptance in science and reflective thought generally, it will be useful to adopt a model of reasonableness. The model I shall assume has the advantage of precise formal articulation. All that I shall have to say in this paper could, however, be recast in other terms. Hence the model is not essential to the argument, as the discerning reader will readily note. I shall assume that it is reasonable to accept a report, hypothesis, theory or other statement if and only if accepting the state ment has a maximum of expected utility where the utilities in question are restricted to those that express our epistemic purposes and preferences. The interesting cases are ones where the statement having a maximum of