Mosquitoes can rapidly develop resistance to insecticides, which is a big problem in malaria control. Current insecticides kill rapidly on contact, but this leads to intense selection for resistance because young adults are killed. Of considerable current interest is the possibility of slowing down or even halting the evolution of resistance. Biologists believe that much weaker selection for resistance can be achieved if insecticides target only old mosquitoes that have already laid most of their eggs. This strategy aims to exploit the fact that most mosquitoes do not live long enough to transmit malaria, due to a long latency stage for the malaria parasite in the mosquito. I will present the results of some mathematical work using stage structured population models that can make predictions about the delayed onset of resistance in the mosquito population when they are subjected to an insecticide that only acts late in life. I will also summarise some ongoing work that includes the malaria disease dynamics and also the consequences of mosquito control using larvicides. Larvae can become resistant to larvicides, but the evolutionary cost of this acquired resistance may be reduced longevity as adults, which reduces the likelihood of the parasites completing their developmental stages and thus can actually benefit malaria control. This is a joint collaboration with Rongsong Liu, Chuncheng Wang and Jianhong Wu.