Evaluation of the edible quality of sweetpotato is being conducted with the traditional sensory evaluation method. However, lack of a standard protocol makes the traditional evaluation method to be subjective and unprecise. To address this problem, this study intended to establish a new sensory evaluation model by using fuzzy mathematics. Firstly, five criteria indices were screened for evaluating the sweetpotato quality, including color, odour, sweetness, stickiness (flour-like taste) and fibrous taste. Each of these criteria indices was seperated and definded for five-point scores. The analytic hierarchy process (AHP) was used to construct a weight set and the industry experts were invited to score each pair of criteria indices of pairwise comparison. The sensory scores of various criteria indices of all tested varieties marked by valuators were used as the rank elements to construct fuzzy relationship matrix. The matrix was built and then used to calculate the result vector of fuzzy comprehensive evaluation. Lastly, the calculation rule of the integrated score of sweetpotato edible quality was formulated and edible quality grading standard of sweetpotato was proposed. To test the accuracy of the new evaluation system, 11 sweetpotato varieties with confirmed quality definitions (evaluating in the sweetpotato of China agricultural research system in 2019) were re-evaluated by the new built evaluation system. The result showed that the fuzzy comprehensive evaluation was more stable than that reveled with the tranditional method. The outcome gained from the new evaluation method is coincidence with the current universal recognition by the sweetpotato of China agricultural research system. Furthermore, the scoring was more accurate because there was a detailed scoring rubric for evaluators to mark. Difference in the tested varieties could be quantified with the integrated score via a fuzzy comprehensive evaluation rather than an expert direct scoring method.