Abstract: In order to explore the key index of burnt-sweet tobacco appearance quality, apriori association rules algorithm was used to analyze the appearance quality of tobacco leaves from Wannan of Anhui province. The results indicated that burnt-sweet tobacco leaves existed strong association degree with maturity, oil, and chroma in a condition of minimum support degree of 0.3 and minimum confidence of 0.5. Meanwhile, the index of color, structure and thickness played important roles for burnt-sweet tobacco leaf, but less than those of the other three indices. Therefore, it is suggested that management system should be established by controlling maturity, oil and chroma during agricultural production and industrial management.