Previous research found that decoy effect varies under different types of products, i.e., consumers’ choice preference for dominating option present unsteadiness with the addition of evoked option. Decoy effect might lose effectiveness in certain situations. As perceived risk has great influence on product evaluating process, it might be an important factor that moderates the decoy effect. In order to explore the deep reasons of decoy effect’s instability, the mechanism of decoy effect is studied from the perspective of product’s perceived risk in this paper. Through the 2(high/low risk)×2(decoy/without decoy) factorial design, 2 types of risk perception product(high/low risk perception product) and 2 levels of decoy situation (original / decoy selection) were examined in this study. In the low risk purchase scenario, participates imaged that they need to buy a toothpaste among the given alternatives independently. Whereas, they imaged themselves purchasing a laptop in the high purchase scenario. The original selection (no decoy condition) has two options, one is the target product which is the dominating alternative, the other one is the competing product. The decoy selection (decoy condition) added an evoked product which is dominated by the target product rather than the competing product. The data were collected from 188 undergraduate and graduate students in a university of china. Cash prizes were used to motivate voluntary participation. There were 4 types of experimental conditions: high risk perception product with no decoy, high risk perception product with decoy, low risk perception product with no decoy and low risk perception product with decoy. Each participant was engaged in both kinds of products and randomly assigned to 1 of the 2 decoy conditions. The final choice, degree of product’s perceived risk and attractiveness of target and competing product were recorded in each purchase scenarios. In addition, the moderating effect of product’s perceived risk was studied in different decoy conditions with multivariate analysis of variance (MANOVA), and the mediating role of relative attractiveness (the difference of the attractiveness of target option and their competing counterparts) was further analyzed with stepwise method. According to our statistics, respondents in the decoy conditions who are lured by the decoy alternative in the low perceived risk product are unlikely to switch their choices. However, they are more likely to switch in their high perceived risk counterparts. The results show that there is an interaction effect of decoy situation and product’s perceived risk, i.e., the influence of decoy product on consumer choice is moderated by product’s perceived risk. The decoy effect arises significantly in high risk perception products, but insignificantly in low risk perception products. On the other hand, it is proved that relative attractiveness of the target product is a fully mediating variable in the model, i.e., product’s perceived risk and decoy situation affect the share of target product by influencing its relative attractiveness. This paper proposed a mediated moderation model to explain the mechanism of decoy effect. Our findings replenish the previous studies of decoy effect and illustrate how firms could take advantage of the context sensitivity of consumers to design more profitable product marketing plans. In summary, high risk perception product is more suitable for the context marketing versus low risk perception product. In other aspects, the target product’s attributes should above the average level among the selection in order to improve the relative attractiveness which is deduced from the mediating effect.