A stochastic frontier model is used to explain the performance evaluation of U.S. university extension providers by organic producers. The model makes explicit the nonmanagerial factors that influence both performance ratings and performance efficiency, defined as achieving a rating as close to the "best" rating as possible. Results indicate that extension agents are performing at relatively high mean efficiency of 0.92, but that the average rating is relatively low at 2.66 on a four-point scale. Several sources of potential bias in ratings are identified as a way for managers to more accurately conduct individual performance assessments. Programmatic changes to emphasize more collaborative research and training are suggested by the positive effects on performance efficiency associated with farmer research commitment and production problem severity.