Many pest repellents are in household use, and a variety of procedures, indices, and statistical tests have been used to examine their effectiveness. However, the effect of action mechanism on these indices has not been adequately discussed. In addition, the interreplication variance of data have not been taken into account in some cases. This study considered the validity of tests used to evaluate repellents and proposed methods to statistically evaluate them. First, various repellents with different mechanisms of action were virtually assumed, and their repelling processes were formulated as differential equations. Numerical and analytical analyses of the formulae revealed that, in some action mechanisms, the indices could be altered by parameters other than the repellent effect and that the indices should be used only for repellents of a specific mechanism. Furthermore, the effect of an actual repellent, by using two different experimental designs and the maize weevil, Sitophilus zeamais Motschulsky (Coleoptera: Curculionidae), were evaluated. This demonstrated that the evaluated effectiveness could vary with experimental design and suggested that the design should reflect the situation in which repellents are actually used. Therefore, statistical methods to quantify the effectiveness of repellents are proposed herein and applied to the results of the above-mentioned experiments. Particular emphasis was placed on the importance of incorporating the interreplication variance in statistical models and estimating confidence intervals. Finally, the validity and limitations of evaluation using indices of repellents' effects, and the identification of important criteria that should be considered when assessing the effects of repellents also are discussed.