To increase marketing efficiency and quality competitiveness in the era of the WTO and FTAs, it is important to know the quality levels of agricultural products that consumers want. The grading standards of Agricultural Products Standardization Management by the National Agricultural Products Quality Management Service is the official quality standards, which, however, are not utilized in the market at all. One of the main reasons behind this is that the existing grades and standards do not reflect the consumer’s preference, but are mostly based on the opinions of scientists and practitioners. This study conducts an experiment to collect data about consumer preference and willingness to pay on important attributes of apple. The quadratic discriminant analysis(QDA) and nonlinear programming(NLP) model are then used to determine the number of grade and the standards for each grade using the optimal grade boundaries. The procedure adopted in this study can be a stepping-stone to the consumer-based grading system.