This study evaluates the spatial genetic structure of baobab (Adansonia digitata) populations from West African agroforestry systems at different geographical scales using AFLP fingerprints. Eleven populations from four countries (Benin, Ghana, Burkina Faso, and Senegal) had comparable levels of genetic diversity, although the two populations in the extreme west (Senegal) had less diversity. Pairwise F(ST) ranged from 0.02 to 0.28 and increased with geographic distance, even at a regional scale. Gene pools detected by Bayesian clustering seem to be a byproduct of the isolation-by-distance pattern rather than representing actual discrete entities. The organization of genetic diversity appears to result essentially from spatially restricted gene flow, with some influences of human seed exchange. Despite the potential for relatively long-distance pollen and seed dispersal by bats within populations, statistically significant spatial genetic structuring within populations (SGS) was detected and gave a mean indirect estimate of neighborhood size of ca. 45. This study demonstrated that relatively high levels of genetic structuring are present in baobab at both large and within-population level, which was unexpected in regard to its dispersal by bats and the influence of human exchange of seeds. Implications of these results for the conservation of baobab populations are discussed.