Sampling of day-resting Anopheles funestus was carried out in September-November 1991, October-December 1992, and November 1994 at two sites near Ouagadougou, Burkina Faso: the small village of Noungou where humans outnumber cattle, and the nearby Fulani settlement of Loumbila where cattle outnumber humans. Collections made inside human dwellings were supplemented in 1992 by outdoor-resting samples from artificial pit-shelters. Indoor-resting An. funestus were also collected in November 1992 and November 1994 in four villages of the Banfora area (southern Burkina Faso) and in a sudanese-sahelian village in northern Burkina Faso (Tougouri). Half-gravid female sub-samples were preserved in carnoy's fixative and processed for polytene chromosome analysis. The material from the two villages near Ouagadougou was analysed by ELISA to know (i) the human/animal origin of the blood meal; (ii) the infectivity for Plasmodium falciparum malaria; and (iii) the possible correlation of these parameters with chromosomal variants. A total of 1416 An. funestus could be scored for the whole polytenic complement, while the origin of the blood meal and circumsporozoite protein (CSP) positivity were asserted from 1076 and 1154 specimens, respectively. With a few exceptions, four polymorphic paracentric inversions (3a, 3b, 2a and 5a in decreasing order of frequency) were observed in all populations. Inversion 2s, whose breaking points include those of inversion 2a, was found only as the heterokaryotype 2s/+ floating at an overall frequency of 3.7% in two villages of the Banfora area and in the two sites near Ouagadougou. Two heterokaryotypes 2a/t out of 186 scored specimens were observed in different years from one village of the Banfora area. Wide variations in inversion frequencies were observed among the samples without consistent geographical or temporal clines. Highly significant departures from Hardy-Weinberg equilibrium were recorded for inversions 3a and 3b in most samples, with the alternative homokaryotypes (standard and inverted) significantly more frequent than expected. Conversely, inversion 5a was in Hardy-Weinberg equilibrium in most samples, whereas the 2a-s inversion system was intermediate between these extremes. However, a deficit of heterokaryotypes was apparent practically in all samples. Significantly higher frequencies of the standard homokaryotypes were recorded (i) in the exophilic samples collected in Loumbila for arrangement 3a; (ii) in the animal-fed sub-sample collected outdoors in Noungou vs. the parallel human-fed sub-sample for arrangements 2a-s, 3a, and 3b, or vs. the samples obtained from indoor catches in both Loumbila and Noungou in the case of inversion 3a; (iii) in the December CSP-negative sub-sample from Loumbila vs. the parallel CSP-positive sub-sample for arrangement 2a. A plausible working hypothesis is that An. funestus in Burkina Faso includes two taxonomic units, one of which is mainly monomorphic standard with most inverted arrangements floating at very low frequencies, and probably uniquely characterised by arrangement 2s, while the other taxon is nearly fixed for arrangement 3ab and polymorphic for all the other inversions at intermediate to high frequencies. The latter would be characterised by a higher vectorial capacity and would probably correspond to An. funestus s.s. from East Africa. About the former hypothetical taxon, its endophily and anthropophily appear less marked and its relationship with other members of the An. funestus subgroup will require specific investigations.