This study demonstrates a tractable method for analyzing schooling investment with risky incomes. Constant relative risk aversion is assumed, and borrowing in a rudimentary capital market is allowed. A linear, variance-components model on log (real income) is estimated. Only unexplained variation is treated as a source of risk. Illustrative empirical results indicate that students should take either 4 years of college or none at all, depending on time preference, loan availability, and degree of risk aversion. Estimate risk-adjusted rates of return to college exceed 10 percent for some parameter values. Risk adjustments for college rates are small but positive.