Various factors that may influence routine and high levels of mosquito infection with cultured Plasmodium falciparum gametocytes are considered in this paper. One of the most important is the choice of an appropriate isolate, with facilities for cryopreservation and a good technique for initiation of cultures. The use of automated culture systems with strict adherence to detail and routine has eliminated much of the variability. The quality of the serum used for the culture of gametocytes and inclusion in the feed material for mosquitoes is of the highest importance. Blood collection for culture purposes must preferably involve alcohol as an antiseptic for cleaning donor skin or suitable receptacles. Mosquito blood meals should not include plasma with citrate phosphate dextrose or sera collected in microtainer tubes or from volunteers on proguanil-chloroquine prophylaxis. Sera of individuals on chloroquine alone do not influence transmission. Haemato-crits of from 5 to 10% permit the culture of equally infective gametocytes. It was impossible to predict the outcome of an infection in mosquitoes based on the number of female gametocytes or gametes. Within any experiment, the oocyst load initially increased, followed by a decline with progressively lower numbers of gametocytes accompanied by a progressive increase in the efficiency of transmission. Some of the variability of mosquito infection within an experiment was due to individual differences in the speed of blood digestion of the mosquitoes. A new membrane feeder is described with three different sizes to accommodate a variety of goals. The new design permitted more mosquitoes to feed within a limited period compared to the older more widely used feeder of equivalent capacity, and mosquitoes feeding on the new-type feeder were frequently better infected than those fed on the older type.