Because grading errors can impede the integrity of the beef grading system, it is important to investigate the existence and possible sources of grader bias. In this article, we use a non-parametric method to identify grader bias in the beef grading system. Using simulation studies, we show that this method performs well without the normal distribution assumption. We find that macroeconomic events and graders’ mood may affect grades using a non-parametric method. Further, the premium/discount analysis in this article suggested that, under the grid price system, livestock producers could have lost financially if human graders had been replaced by camera-grading system during our sample period.