Abstract: In order to establish a highly effective determination method of starch in flue-cured tobacco, the determination method of tobacco starch using ultrasonic extraction and iodine colorimetry-dual wavelength spectrophotometry was optimized using Yunyan 85 as the raw material. The effects of ultrasonic time, ultrasonic power, perchloric acid concentration and liquid-solid ratio on the ultrasound-assisted extraction of amylose and amylopectin in tobacco were investigated. The extraction process was optimized using the response surface Box-Benhnken experimental design principles and the response surface methodology. The results showed that the optimal extraction conditions were that with the perchloric acid mass concentration of 49.0%, the ultrasonic power of 400 W, the ultrasonic time of 8.50 min, and the liquid-solid ratio of 27 mL/g. At the optimized conditions, the recovery of amylose was from 104.45% to 107.48%, recovery of amylopectin was from 96.42% to 100.51%, and the RSD(n=3)was less than 3.5%. This suggested that the optimized determination method of starch used by the Box-behnken design and response surface analysis have good precision and accuracy and are suitable for the detection of starch in flue-cured tobacco.