A variety of crop revenue insurance programs have recently been introduced. A critical component of revenue insurance contracts is quantifying the risk associated with stochastic prices. Forward-looking, market-based measures of price risk which are often available in form of options premia are preferable. Because such measures are not available for every crop, some current revenue insurance programs alternatively utilize historical price data to construct measures of price risk. This study evaluates the distributional implications of alternative methods for estimating price risk and deriving insurance premium rates. A variety of specification tests are employed to evaluate distributional assumptions. Conditional heteroskedasticity models are used to determine the extent to which price distributions may be characterized by nonconstant variances. In addition, these models are used to identify variables which may be used for conditioning distributions for rating purposes. Discrete mixtures of normals provide flexible parametric specifications capable of recognizing the skewness and kurtosis present in commodity prices