First Paraghraph: A method is given for quantitative analysis of wine headspace. The reproducibility of the method, for 24 identified peaks, is _+10-15% if internal standards are chosen correctly. Unidentified peaks of lesser intensity generally have larger coefficients of variation. Variations in headspace analysis between field replications were too large for the data to be pooled. All calculations were done on individual field replicates. Stepwise discriminant analysis was used to show that the wines could be correctly sorted on the basis of analysis of either must, wine, or headspace. Selective must, wine, and headspace data were correlated linearly with crop levels.