This paper investigates the implications for asset price dispersion of conventional security valuation models. Successively sharper variance bounds on asset prices are derived. Large-sample tests of the bounds are determined and applied to aggregated and disaggregated price and earnings data in U.S. corporations.