This paper uses stochastic production frontiers (SPF) and stochastic distance frontiers (SDF) to measure technical efficiency for a sample of dairy farms in Abasto Sur, Argentina. The data is a highly unbalanced panel including 46 farmers from 1997/98 to 2001/02 with a total of 82 observations. Four alternative models based on the Battese and Coelli framework are evaluated. Average technical efficiency across the four models ranges from 67.2% to 88.4% while the correlation coefficient for technical efficiency scores ranges from 0.632 to 0.976. There is significant technological regress, with an annual average of 16.8% and 17.7% for the two “best” models. The results indicate that alternative specifications can have a significant impact on the results, specifically on mean efficiency figures.