This paper critically examines appropriateness of the nonstandard profit function (NSPF) as a tool for measuring profit efficiency. We show that there are some fundamental problems with the NSPF approach, the way it is currently specified and estimated. This is true irrespective of whether technical inefficiency is included or not. Instead of estimating profit efficiency directly from the NSPF, we suggest alternative specifications and recommend using the cost function approach and deriving profit efficiency therefrom. Data from US banks are used to estimate the cost function and the NSPF. Results from alternative specifications are then compared and analyzed.