This study investigates the relationships between farm size, milk yield, cost of production, and technical efficiency in the Alberta dairy industry. Estimates of a stochastic production frontier are obtained with two alternative methods; an iterative "average frontier: (AF) procedure and a maximum-likelihood composed error (CE) term method. An index of technical efficiency is calculated for every herd in the sample, with the AF method resulting in an average efficiency ratio of 85 percent, and the CE method producing an average efficiency ratio of 83 percent. Regressions of production cost on milk output, herd size, and efficiency are used to test for the effects of size economies, yield economies, and technical efficiency on production cost. These results suggest that herd expansion, on average, would lower the average cost of production throughout the province. Romain and Lambert use a similar method in a study of Quebec and Ontario dairy farmers which shows a limited potential to exploit economies of herd size. While not a formal test of the similarity of the two industries, the results of this study indicate a significant difference between the optimal structure of dairy production in Alberta and Quebec. Such regional differences will have important implications for the possible reapportionment of the national milk market, whether by regulatory or free-market mechanisms.