Batesian mimicry is a phenomenon where a palatable species protects themselves from predators by resembling appearance of noxious species. In Papilioniadae butterflies, there is a system where only females mimic, and both mimetic females and nonmimetic females coexist in the same species. Two major hypotheses have been proposed to explain the evolutionary paradox of why nonmimetic females are not eliminated despite their disadvantage in predation avoidance. First, the frequency of mimetic females depends on the relative frequency of unpalatable models, and second, is that there are some costs in mimetic females which offset the benefit of mimicry. However, empirical data testing these hypotheses are still insufficient primarily due to difficulty in evaluating predation pressure in the wild. In this study, we aimed to evaluate predation pressure in the field by comparing the frequency of mimetic genotypes between males and females in Papilio polytes, a butterfly showing female-limited Batesian mimetic polymorphism. If only negative frequency-dependent selection is at work, the mimetic genotype frequency should be similar between males and females, indicating that the predation pressure on the two types of females is balanced. On the other hand, if a mimicry trade-off is at work, the mimetic genotype frequency should differ between the sexes, being higher in females, which might indicate biased predation pressure on nonmimetic females. The surveys on five islands (Okinawa, Miyako, Ishigaki, Taketomi, and Kikai) with different frequencies of toxic models during 2022–2024 showed no significant difference in the frequency of mimetic genotypes between sexes on any island. The ordinal logistic regression analysis pooling the data of all islands showed that only the relative frequency of toxic models had a significant positive effect on the mimetic genotype frequency. In addition, genotype frequencies were consistent with Hardy-Weinberg equilibrium with some exceptions. Overall, these results supported the frequency-dependent selection hypothesis better. However, the results of some islands were inconsistent with the prediction. In particular, the female mimetic genotype frequency on Ishigaki has drastically decreased compared to the data collected before (2014–2017). This suggests that the mimetic genotype frequency sometimes becomes out of equilibrium, fluctuating over several years. Theoretical predictions that do not necessarily assume an equilibrium of mimetic genotype frequency may be needed. Either way, assessing the genotype frequency of P. polytes in the wild would be useful to understand its ecology and evolution of Batesian mimicry.