Two models for assessing the performance of herbivores on a variety of host plants are developed by combining knowledge of population genetics and population ecology, especially that of the fixation probability of mutant genes. The absolute host performance model precisely assesses host performance for one herbivore population based on parameters of fecundity, larval survivorship, and selection pressure. The relative host performance model compares host performance for one population among different host plant species and for several populations on the same host species. Two herbivore populations, Bemisia tabaci (Gennadius) and Tetranychus truncates Ehara, were used to validate the absolute and relative host performance models. Results indicated that the assessment systems of host performance were reasonable and reliable. These models could be applied to a wide range of herbivore species for assessing their performance on host plants.