The demand for wine is generally estimated on an aggregate level as a single commodity. However, as recent history shows us, the demand for wine not only varies considerably by varietal, but also by price point within each varietal. As a result, although estimates of the demand for wine may be beneficial to the wine industry as a whole, they provide little benefit to individual wine producers. Using scan data of purchases from US retail chain stores, this paper uses store keeping unit (sku) level data to overcome the limitations of prior research on the demand for wine by providing estimates for the demand for wine by varietal and price point. We also provide estimates of own price effects, income effects as well as cross price effects by color, varietal and price point. Problems of endogeneity inherent in demand estimation are corrected by utilizing a novel instrumental variable technique using grape prices as the instrument.