Diapause, a form of dormancy to delay or halt the reproductive development during unfavourable seasons, has evolved in many insect species. One example is aestivation, a summer adult-stage diapause, enhancing malaria vectors’ survival during the unfavourable dry season (DS) and their re-establishment in the next rainy season (RS). This work develops a novel genetic approach to estimate the number or proportion of individuals undergoing diapause, as well as the breeding sizes of the two seasons, using signals from temporal allele frequency dynamics. Using Anopheles coluzzii as an example, our modelling shows the magnitude of drift is dampened at early RS when previously aestivating individuals reappear. Aestivation severely biases the temporal effective population size (N_e), leading to overestimation of the DS breeding size by 1/(1-α)^2 across one year, where α is the aestivating proportion. We find sampling breeding individuals in three consecutive seasons starting from a RS is sufficient for parameter estimation, and perform extensive simulations to verify our derivations. This method does not require sampling individuals in the dormant state, the biggest challenge in most studies. We apply the method to a published An. coluzzii dataset from Thierola, Mali, and the estimated aestivating proportions were 39%-79%. These results will inform the development of genetic approaches to vector control. Beyond mosquitoes, our method and the expected evolutionary implications are applicable to any species in which a fraction of the population diapauses for more than one generation, and are difficult or impossible to sample during that stage.