A central assumption of models of risky decision-making is that value and subjective probability contribute independently to decisions. A number of studies have tested this assumption with a variety of experimental designs including asking subjects to make direct estimates of probability, guesses, wagers, and ratings of gambles. The results of these suggest that there is evidence of interaction, particularly when subjects choose between events of equal probability, one of which is valued. Theories of the conditions under which interaction occurs were considered, and there is support for the importance of distinctions between chance and skill control of outcomes, simple and complex tasks, and w~l l -and ill-defined tasks.When subjects make judgements of the likelihood that some event will occur, it is often the case that the occurrence of the event will have value for them, or that they will prefer the cccurrence of one event rather than another. The question of the independence of the value of an outcome and the subjective probability of attaining it is of considerable theoretical importance in the psychology of risky decision-making.win outcome A with probability P or lose outcome B with probability Q, in order to make a bid for the gamble or to choose between it and another gamble similar in form. There are three principal approaches to understanding behaviour at this task, and the independence of value and subjective probability is an important question for each of them. The Subjective Expected Utility (SEU) model states that when certain axioms, including the independence of utility and subjective probability, are satisfied, utility and subjective probability functions can be constructed to account for decisions. Tests of the model have either examined the descriptive validity of the axioms or attempted to fit the model to choice data, and there is considerable evidence of behaviour incompatible with the model's axioms (Slovic et al., 1977). Evidence that subjective probability was not independent of value would cast further doubt on the descriptive validity of the model. decision-maker integrates the four items of information, A , B, P, Q, into a judgement, through addition and multiplication operations, and that this integration takes place at the subjective level, i.e. Judgement = w ( P ) . $(A)+ w(Q). s(B), where the ware subjective weights corresponding to probabilities, and the s are subjective scale values corresponding to the outcomes. Tests of the model depend upon analysis of variance techniques, and significant interaction terms would reject the model. The interactions of interest here would be those between outcomes and probabilities.The Information Processing Model (Slovic & Lichtenstein, 1%8) considers that subjects' evaluations of a gamble may be described by the equation , Evaluation = w,P+ w,A + w3Q+ w,B+ u, where the weights, w, reflect the importance of each of the four dimensions for the subject in determining the attractiveness of the gamble, and are in practice measured by the size of co...