This paper estimates technical efficiency (TE) measures using four alternative production frontier models, and evaluates the sensitivity of the results to the choice of methodology. The Cobb‐Douglas functional form along with data for 1982 and 1983 from 404 dairy farms located throughout six northeastern states in the US are used in the estimation. A general conclusion is that, broadly speaking, frontier function models are neutrally upwardly scaled versions of the OLS or average model. A second conclusion is that different models yield markedly different efficiency levels across firms. However, the correlation between the indexes from the various methods is high, which implies that the ordinal ranking of firms according to their measured level of technical efficiency appears to be independent of the method used for a given year. By comparison, the correlation between efficiency indexes for the same method across time, although positive, is much lower than the previous set of correlations. Correlation analysis of efficiency versus farm size and of efficiency versus returns over variable costs, based on the alternative models, yielded consistent results.