According to FAOSTAT 2014, the production of potato is huge in China, which was almost 96.1363Mt/a, accounting for 24.97% of global production, but subsequently lots of potato stem leaves were produced and had not been effectively utilized. At the same time, the corn straw is also a kind of huge reserve of rural biomass energy, of which the production was about 44Mt/a. The potato planting area in China reached 5.647Mhm2. Southwest mountainous area, northwest area, Inner Mongolia and northeast area are the main planting areas of potatoes, of which the cultivated area approximately accounted for 62% of total potato planting area in China. Thus, large-scale biogas project in these areas can be built by using potato stem leaf as a kind of raw material for anaerobic digestion, which is an efficient path for potato stem leaf recycling. Therefore, based on the early single factor experiment, the parameters optimization of anaerobic digestion process which used potato stem leaf mixed with corn straw as raw material was studied. The quadratic regression orthogonal rotational combination design was used to determine the influences of material concentration, carbon and nitrogen ratio and inoculum quality percentage on mesophilic anaerobic digestion of potato stem leaf mixed with corn straw. Through the research, the quadratic regression model was optimized and decreasedimension was analyzed by the software of Design-Expert. The results showed that the optimum conditions were confirmed as material concentration of 9.87%, carbon and nitrogen ratio of 18.73 and inoculum quality percentage of 20.82%. The predicted TS gas production was 54.55mL/g. In addition, the primary sequence of effect of three factors was inoculum quality percentage, material concentration and carbon and nitrogen ratio. Furthermore, the reciprocal action between inoculum quality percentage and material concentration was obvious. During the verification test, the relative error of TS gas production between model calculation value and experimental value was below 5%. The variance analysis was not significant, and the model matched well with the experiment results.