A longitudinal study based on mosquitoes sampled by larvae prospecting and adult catches on humans was carried out during 1999 in the rice field area of the Kou Valley (South-West Burkina Faso) to evaluate the malaria transmission level. Two sites were studied: VK5 located in the rice field centre and VK7 in the periphery. Irrigation is sub-permanent and two crops are grown each year: from February to June during the dry season and from July to November during the rainy season. A man sleeping in VK5 without any protection is exposed to more than 60,000 mosquito bites/year. Two majors vectors are present: Anopheles gambiae sl. and An. funestus. The An. gambiae M form which breeds in rice fields constitutes the majority of the complex. At the periphery of the rice fields, we observed a mix of M and S forms during the rainy season, the latter form coming from classical breeding sites created by rainfall (residual puddles). An. arabiensis is rare in this environment while we can find it in sympathy with An. gambiae in the surrounding savannah. An. gambiae remains present throughout the year. Its dynamics depends closely on both season and rice cycle. There are two density peaks, in March (210 bites/man/night) and in July (306 b/m/n), corresponding to the moment when rice is planted. The latter peak is more important due to additional productivity of residual puddles created by rainfall. These large and very young populations can not transmit malaria. Growing rice reduces larvae productivity: transmission occurs when the adult population decreases and becomes older. Variations in parity rate and sporozoitic index are inversely related to mosquito density. Finally, the An. gambiae population is lowest during the dry season in January when irrigation stops. An. funestus does not develop in rice fields but rather in drains made of dirt, in addition to the classical natural breeding sites at the end of the rainy season. The density of An. funestus is 25 to 30 times smaller than that of A. gambiae, as observed during 8 months in VK5 and 10 months in VK7, with a peak of 16 b/m/n in January. Its mean parity rate and sporozoitic index is higher than those of An. gambiae. An. funestus plays an additional role in malaria transmission at the same period as An. gambiae but more shortly and less intensively. Finally, malaria transmission is due mainly to An. gambiae (90%) and, to a lesser extent, to An. funestus (10%). It occurs in VK5 during two periods of four month search : from May to August and from November to February. In VK7, transmission is also bimodal but it occurs more extensively throughout the year with only July and Septemberas exemptions of transmission. The annual inoculation rate amounts to 697 and 515 infected bites per man. This value is higher than those obtained in 1984/1985 : 50 and 60 infected bites per man per year. This increase is due to both the growth of aggressive vectorial density and the sporozoitic index. Growth of adult density can be explained by the environmental modification and/or better adaptation of vectorial species. The sporozoitic index of vectors has been obtained using Elisa, which is twice as sensitive as the microscopic observations used in 1984/1985. This can partly explain the increase in the sporozoitic index. Other hypotheses in relation to human parasitological inputs have to be evaluated more accurately. The level and rhythm of transmission in the rice field in 1999 are more important than those registered in the surrounding savannah the same year. However, this important transmission is not proportionally related to the high vectorial density found in rice fields. Probably, it does not imply more malaria diseases, but its intensity and rhythm could concentrate malaria cases in young people with an early acquisition of premunition. Pyrethroid impregnated bednets should be used to prevent malaria especially since high mosquito nuisance drives local populations to use it and since the An. gambiae M form is particularly susceptible to this class of insecticides.