Based on a longitudinal survey performed in 1982–1983, a vector control strategy was implemented from 1985 onwards in a malarial-dense area of Burundi. One annual round of indoor spraying with malathion greatly reduced both the parasite load and the parasite rate in the population until 1989. However, from 1990 to 1993, a progressive resurgence of malaria was observed in most villages. For the present study, two villages were selected on the basis of their differential response to house spraying. In the village of Mulira surrounded by rice fields, the excellent results observed in the past have been followed by recent increases in parasite rates. In the village of Murengeza, also located in the rice growing area but near a river, the spraying had less impact. The inoculation rate was found to be similar in both villages, but the transmission peak occurred at the end of April in Mulira, and two months earlier in Murengeza. Indoor spraying with lambdacyhalothrin was carried out on 26 April 1993, one month too late according to the strategy intended. As no sporozoite mosquitoes were observed during the six months following spraying, this strategy should be maintained but, in villages near rivers, the application should commence much earlier, in mid-January. Anopheles gambiae Giles sensu stricto (Diptera: Culicidae) and A. funestus Giles were found to be very endophilic species, whereas the dominant A. arabiensis. Patton was highly exophilic. Therefore it is recommended that treatments should not only be applied to human dwellings but also to other structures such as animal sheds, kitchens, etc, shown by earlier studies to be resting sites of A. arabiensis. This study underlines the need for regular reassessment in vector control programmes.