Under the Federal Milk Marketing Order (FMMO) and California milk pricing systems, minimum milk values are determined by the value of components important in dairy product manufacturing. This implies that milk values will vary across farms due to different solids composition (fat, protein and other solids) delivering to the same processing plant. Milk composition can be managed by farm operators by a number of management activities such as: breed choice, number of lactations to keep a cow in the milking herd, ration formulations, feeding management, whether to milk 2 or 3 times daily, and cow comfort. Given the milk pricing systems used for a majority of raw milk in the U.S., dairy farm operators are faced with an environment of maximize profits via a multi-output production function, (i.e., production of milk components). Previous analysis of dairy farm efficiency has typically used the total amount of milk produced (cwt of lbs) as a measure of output, not the production of its components. In this paper, we use hedonic aggregation functions to generate output indices in the evaluation of an input-oriented distance function. We use data from the 2005 USDA Agricultural Resource Management Survey (ARMS)-Dairy Survey for this analysis. A unique feature of the 2005 survey is that it contains information on the annual amount of milkfat and protein produced by the milking herd. From this analysis we find that the estimated technical efficiency when using component amounts as output measures has less variance, but larger range, compared to the method using milk yield as output. A majority of dairy operations generate technical efficiency measures of more than 9.0 (with 1.00 being the possible maximum.