I. INTRODUCTION IN A RECENT paper [2] Peter Fortune has provided us with an excellent theoretical and empirical development of the determinants of optimal life insurance purchases. However, the theoretical comparative statics results are frequently indeterminate, sometimes at crucial points. Fortune's analysis develops the uncertain choice problem in terms of statistical parameters; means and variances. By contrast, the present paper will adopt a state-preference approach to the problem.1 This approach is capable of resolving the theoretical ambiguities of Fortune's analysis and should provide a better background for the empirical work. In a simple two state model the advantages of the state-preference approach are palpable. First, the analysis is easier both graphically and mathematically than an analysis based on means and variances. More importantly, in concentrating on statistical parameters the mean-variance approach may discard useful information. A state-preference approach will not. Indeed, the ability of the present analysis to resolve the ambiguities alluded to earlier stems precisely from the fact that more information is used as an input to the analysis. The result is more output. Probably the most surprising ambiguity in Fortune's analysis concerns the impact of a rise in anticipated wage income on the quantity demanded of life insurance (Theorem 5). We will show that the only requirement necessary for life insurance demand to be an increasing function of wage income is that claims to consumption in the state 'breadwinner dies' be superior goods. This is a result consistent with Fortune's empirical work and is demonstrated in Section III. Two additional points are examined. In Section II it is shown that the insurance contract which Fortune examines is unlikely to be an optimal one from the point of view of the individual purchaser. This is a minor point but its development will simplify the later analysis. Finally, in Section IV it is shown that the ambiguity surrounding the relationship between the price of insurance and the quantity demanded (Theorem 6) can also be resolved. Again, all that is required is that claims to consumption in the state 'breadwinner dies' be superior goods. It is hoped that these results will strengthen Fortune's analysis. II. THE NATURE OF THE OPTIMAL INSURANCE CONTRACT This section deals with a minor point, but one which can be made rather quickly and will simplify the analysis of the next two sections. In Fortune's work, W1 represents nonhuman wealth at the beginning of period 1, and W2 is the value of nonhuman wealth at the beginning of period 1 plus the present value of human wealth. The difference between them is wage income (properly dis* Associate Professor of Economics, Indiana University. The final version of this paper has benefited from the comments of Peter Fortune.