A vast body of literature exists on estimating hedonic price functions which relate the price of wine to its attributes. Some studies have employed producer specific variables such as quantity sold and producer reputation in hedonic functions to potentially capture supply influences on prices. This paper recognizes that the original Rosen (1974) hedonic theoretic framework excludes producer specific variables from the hedonic price function and justifies their inclusion only for second-stage attribute supply estimation. We use the two-stage Rosen approach employing data from multi-markets for the same wines to identify supply functions. The application to Australian produced wines demonstrates the importance of a wine's quality and age as attributes in inverse supply functions. Counter to expectations a direct relation between producer size and marginal attribute costs is estimated which appears to be due to the method employed rather than the peculiarities of the data.