Models of malaria epidemiology and evolution are frequently based on the assumption that vector-parasitic associations are benign. Implicit in this assumption is the supposition that all Plasmodium parasites have an equal and neutral effect on vector survival, and thus that there is no parasite genetic variation for vector virulence. While some data support the assumption of avirulence, there has been no examination of the impact of parasite genetic diversity. We conducted a laboratory study with the rodent malaria parasite, Plasmodium chabaudi and the vector, Anopheles stephensi, to determine whether mosquito mortality varied with parasite genotype (CR and ER clones), infection diversity (single versus mixed genotype) and nutrient availability. Vector mortality varied significantly between parasite genotypes, but the rank order of virulence depended on environmental conditions. In standard conditions, mixed genotype infections were the most virulent but when glucose water was limited, mortality was highest in mosquitoes infected with CR. These genotype-by-environment interactions were repeatable across two experiments and could not be explained by variation in anaemia, gametocytaemia, blood meal size, mosquito body size, infection rate or oocyst burden. Variation in the genetic and environmental determinants of virulence may explain conflicting accounts of Plasmodium pathogenicity to mosquitoes in the malaria literature.