The continual recruitment of new individuals makes it difficult to study both the survival of multivoltine mosquitoes, and the size of the infectious reservoir in narural populations of malaria vectors. During long-term surveillance of a population of Anopheles gambiae Giles sensu lato in a Tanzanian village by daily light trapping, a temporary dry spell resulted in the cessation of recruitment for a period of 33 days, and a decline in numbers of A. arabiensis Patton caught from over 2000 to less than 10 in a sentinel house. Traps placed elsewhere in the village indicated similar proportionate declines although numbers caught varied according to location. A survival rate of 83% per day was estimated from the rate of population decline. Survival was unrelated to the size of the mosquitoes. The infectious reservoir (the chance of a mosquito acquiring an infection) was estimated to be 2% per feed. The exploitation of fortuitous events which temporarily eliminate a single stage in the life cycle has general applicability in the study of the bionomics of multivoltine insects.