The research is concerned with assessment of utility functions for multi-numeraire consequences. More specifically, it is proven that given von Neumann and Morganstern's 'axioms of rational behavior' and two additional assumptions, the utility function for (x sub i, y sub i) consequences must be of the form U sub xy(x sub i, y sub i) = U sub x(x sub i) + U sub y(y sub i) + K U sub x(x sub i) U sub y(y sub i). K is a constant that must be empirically evaluated. It is shown that this form, known as a quasi-separable utility function, is more general than the separable utility function and nearly as easy to use. The implications and ramifications of such a utility function and its requisite assumptions are discussed in detail. Expressions for evaluating the expected utility of a probabilistic consequence are derived. The problems and technique of practical application of the theory are considered. A discussion of the usefulness of this work and of possible future research topics concludes the report. (Author).

A continuous complete preference ordering is defined on an arcconnected, topologically separable product space S = S1xS2x.xSn. Call 1,.n sectors and say that a set A of sectors is separable if the conditional ordering on A, given what happens off it, is independent of the latter, essential if it matters what happens on A, at least sometimes, and strictly essential if it always does. It is shown how to determine the structure of the utility function, given a collection of separable sets, when each sector is strictly essential. Various examples are discussed, an alternative approach sketched, and, finally, the requirement of strict essentiality replaced by the basically nugatory condition that each sector be essential. The results can, of course, be applied to other functions, too. (Author).