A simple valuation model that allows for time variation in investment opportunities is developed and estimated. The model assumes that the investment opportunity set is completely described by two state variables, the real interest rate and the maximum Sharpe ratio, which follow correlated Ornstein- Uhlenbeck processes. The model parameters and time series of the state variables are estimated using data on US Treasury bond yields and expected in- flation for the period January 1952 to December 2000, and, as predicted, the estimated maximum Sharpe ratio is shown to be related to the equity premium. In cross-sectional asset pricing tests using the 25 Fama-French size and book-to-market portfolios, both state variables are found to have significant risk premia, which is consistent with the ICAPM of Merton (1973). In contrast to the CAPM and the Fama-French 3-factor model, the simple ICAPM is not rejected by cross-sectional tests using the 25 Fama-French size and B/M sorted portfolios. Returns on the 30 industrial portfolios do not discriminate clearly between the three models. When both sets of portfolios are included as test assets all three models are rejected, but the estimated risk premia for both ICAPM state variables are significant while those associated with the Fama-French arbitrage portfolios are insignificant.