In an area of holoendemic malaria in Northern Tanzania, Anopheles gambiae s.l. females were age-graded by Polovodova's method and dissected for sporozoites. Age-specific sporozoite rates implied that mosquitoes acquired new infections at all ages. The extrinsic period lasted just over 3 gonotrophic cycles (9-11 days). Very high sporozoite rates in the oldest females implied the absence or rarity of genetic refractoriness to infection. A method is described for estimating the proportion of bloodmeals which result in mosquito infection. This method makes relatively few assumptions about mosquito behaviour, and could be useful for evaluating transmission-blocking interventions. Overall, it is estimated that about 21% of meals are infectious. This is much higher than previous estimates derived either from experimental mosquito feeding studies or from similar age-grading data collected from the same area in 1962. Various alternative explanations are considered, and it is concluded that there has been a 2.5-fold increase in human infectiousness in the last 25 years. This is partly attributable to suppression of human infectiousness by widespread chloroquine usage during the 1960s, followed by removal of this effect by drug resistance. It is argued that chloroquine would be expected to select for increased infectivity in the parasite, and this may also have contributed to the observed increase.