A model is developed to explain Florida citrus planting levels by variety. The varietal choice is based on the expected prices and price variances/covariances of the varieties under consideration. Overall planting returns are maximized for a given level of price risk. The model's price coefficients are similar to those of the Theil and Barten Rotterdam demand model. As in the Rotterdam model, both absolute and relative price coefficient specifications are considered, allowing an examination of restrictions related to the price risk. The empirical analysis considers two restricted specifications -- a varietal independence model, based on the assumption that only the price variances are important for predicting planting levels, and a group independence model that additionally assumes the price covariance for two varieties from the same (different) group(s) is nonzero (zero). Both models substantially reduce the parameter space and may be of interest in situations where lack of degrees of freedom is an issue. A varying parameter model was also estimated, indicating that with growth in the world orange-juice market, orange prices have been perceived to be less unstable, and orange planting responses to changes in total citrus acreage planted and orange prices have been stronger.