This paper examines the responsiveness of investment to q (i.e., the ratio of a firm's market value to the replacement cost of its assets) using data on a unique type of firm: Real Estate Investment Trusts (REITs). For REITs, we have high quality estimates of the net asset value of the firm that we use to create relatively accurate measures of Tobin's q. In addition, REITs have institutional features that mitigate some of the complications faced by previous studies. We have three main results. First, there is little evidence of a statistical link between REIT investment and a traditional accounting-based measure of q. Second, REIT investment is highly sensitive to estimates of q that are based on analysts' appraisals of asset value. A REIT whose NAV-based q ratio rises from 1.0 to 1.1 will increase its assets by 4.3 percent in the next year. Third, the difference between the appraisal-based measure of q and the traditional accounting based measure typically increases with the age of the firm's assets and varies across types of properties. These results suggest that measurement error in q can lead to appreciable downward biases in investment sensitivities, even in an industry that seems to meet many of the assumptions in Tobin's original paper, but that Tobin's investment model performs well with a better measure of q.