Rising numbers of malaria cases and deaths underscore the need for new interventions. Long-acting injectable medications, such as those now in use for HIV prophylaxis, offer the prospect of a malaria chemical vaccine, combining the efficacy of a drug (like atovaquone) with the durability of a biological vaccine. Of concern, however, is the possible selection and transmission of drug-resistant parasites. We addressed this question by generating clinically relevant, highly atovaquone-resistant, Plasmodium falciparum mutants competent to infect mosquitoes. Isogenic paired strains, that differ only by a single Y268S mutation in cytochrome b, were evaluated in parallel in southeast Asian (Anopheles stephensi) or African (Anopheles gambiae) mosquitoes, and thence in humanized mice. Fitness costs of the mutation were evident along the lifecycle, in asexual parasite growth in vitro and in a progressive loss of parasites in the mosquito. In numerous independent experiments, microscopic exam of salivary glands from hundreds of mosquitoes failed to detect even one Y268S sporozoite, a defect not rescued by coinfection with wild type parasites. Furthermore, despite uniformly successful transmission of wild type parasites from An. stephensi to FRG NOD huHep mice bearing human hepatocytes and erythrocytes, multiple attempts with Y268S-fed mosquitoes failed: there was no evidence of parasites in mouse tissues by microscopy, in vitro culture, or PCR. These studies confirm a severe-to-lethal fitness cost of clinically relevant atovaquone-resistant P. falciparum in the mosquito, and they significantly lessen the likelihood of their transmission in the field.