Abstract: Quantitative structure-activity relationship (QSAR) model of mosquito repellent plays an important role in the synthesis of new mosquito repellent and interpretation on the mechanism of repellent. In this study, 40 amides were evaluated and effective protection time was used as the indicator of their activities in Aedes aegypti. For each compound, 1 773 molecular descriptors were obtained using PCLIENT(http://www.vcclab.org/lab/pclient/start.html). Finally, 8 descriptors were obtained after screening using binary matrix shuffling filter (BMSF) and worst descriptor elimination multi-round (WDEM). High-precision nonlinear QSAR model (F=8 465,R2=0.999 6) was built based on Support Vector Regression (SVR). SVR interpretability system analysis showed that there was an obvious nonlinear relationship between retained molecular descriptors and the activity of amide mosquito repellent. Topological polar molecular surface area (TPSA (Tot))was the most important molecular descriptors for mosquito repellent, the smaller the value,the higher the activity. Electronegativity significantly influenced repellent activity, the strength of its value,the higher the activity.