Anopheles implexus (Theo.) was studied at the Zika Forest near Entebbe, Uganda, from 1966 to 1971, with central reference to biting activities and host selection. Seasonal fluctuations of populations appeared to be correlated with lake levels, not with local rainfall. Results of biting, resting and nectar-feeding catches at various points across the forest floor indicated that diel activity patterns of man-biting were related primarily to the proximity of resting female A. implexus and only secondarily to the influence of environmental variables. Latent effects of moon phase on day-time man-biting activity patterns support this view. Blood-meal identifications showed that the great majority of blood-fed females resting in forest had fed on cattle, indicating that they left the forest at night. This was confirmed by direct catching from cattle outside forest, while the virtual absence of A. implexus resting near these bait animals, together with the fact that man is very seldom attacked away from forest, indicated that different kinds of host selection strategy were involved. The diel activity pattern of biting on cattle taken into forest was then found to be very different from that on man. Age-grading studies helped to substantiate these differences. Hourly sampling of A. implexus arriving at representative resting sites in forest indicated an exodus of most of the adult mosquito population from forest at dusk and a return shortly after sunrise. The two basic host selection strategies evidently involved (a) active or (in effect) ‘hunting’ flight, mainly outside forest with a tendency to restrict choice to most strongly attractive hosts, and (b) passive or opportunistic attack from rest, in which some of the unfed females sheltering may be activated by the proximity of a host which need not be strongly attractive. The latter situation was investigated further through analysis of short-term day-time patterns of attack on man in forest which are postulated to be a function of mosquito response delay (involving both resident and incoming elements of the sheltering population), which in turn determines depletion rates (involving only the resident element available at the time of the host's arrival). In A. implexus the response delay appears to vary according to time of day. Anomalous results are finally re-considered and it is concluded that the different patterns of active and passive host selection could arise from a single circadian rhythm of responsiveness, with flight characteristics and day-time inhibition modulating responses to specific and non-specific host stimuli in different ways.