This paper develops a method of estimating the coefficient of relative risk aversion (g) from data on labor supply. The main result is that existing estimates of labor supply elasticities place a tight bound on g, without any assumptions beyond those of expected utility theory. It is shown that...