Grading systems are often introduced to address the classic adverse selection problem associated with asymmetric information about product quality. However, grades are rarely measured perfectly, and adverse selection outcomes may persist due to grading error. We study the effects of errors in grading, focusing on asymmetric grading errors- namely when low-quality product can erroneously be classified as high quality, but not vice versa. In conceptual model, we show the effects of asymmetric grading errors on returns to producers. Application to the California prune industry shows that grading errors reduce incentives to produce more valuable, larger prunes.