Humor is widely used in advertisements, However it is not applicable to all. If not used properly, humor would reverse the influence the advertisement intends to exert. This research tried to find out to which type of merchandise humor is more applicable, using Product Color Matrix as the standard classification for merchandises. Two hypotheses were put forward: H1, Humor always harvested good advertising effect; H2, Humor acted differently on merchandise of different type. 16 rearranged advertisements were shown to 160 college students, who were randomly picked to take the questionnaire. In this experiment of Balanced Design, Univariate Analysis of Variance and multivariate analysis of variance were applied to process the datas, among which the fact whether humor is used in advertisement was independent variable, evaluation and order of the ads were covariates, impression, preference and purchase intention toward the merchandise were dependent variables. Research results shows: Merchandises classified as big tools, little treats, little tolls gained a better image while humor was applied in their advertisements. The images of merchandises classified as big toys would not be affected either humor is applied. Questionnaire takers preferred ads without humor to that with humor when it comes to big toys. Big toy's ads without humor triggers more purchase intention than ads with humor, and the situation was on the contrary when it comes to little tolls. To sum up, humor is more applicable to ads of merchandise that are low risk and functional products, not which are high risk and hedonic ones.