From all reports, expected utility theory is dead. The reports are greatly exaggerated. This study makes two modifications which revive expected utility theory. Rather than directly modelling risk preferences by a von Neumann-Morgenstern utility function of wealth, risk preferences and the expected utility of wealth are derived from consumption and investment decisions over time. Rather than using future wealth as the reference point for evaluation risk preferences, current wealth is used instead. The revived theory is both normative and descriptive. It specifies how rational people ought to make decisions under risk and explains the major empirical findings about how people actually make decisions. For example the Allais Paradox and its variations, preference reversals and framing effects all result from rational decisions by uniformly risk-averse people. Moreover, apparently risk-seeking behaviour can result from risk-averse people with low rates of time preference taking risks to save for the future. The revived theory also shows why eliciting certainty equivalents cannot measure peoples' risk preferences but leads to new procedures for measuring both time and risk preferences. (JEL D81, D91).