complement an experimental study of the possible interest for plant breeders of an allele increasing recombination rates. In the first simulation study, 4 populations of 10 individuals were simulated in the case of high (45%), medium (19.6%) and low (6.9%) recombination frequencies. Ten-plant populations were derived by one generation, either from selfing or from crossing in a circular mating design, from each of the 4 populations. For each initial population, the distribution of the genetic variances in the populations obtained by one mating system overlapped widely in the 3 cases of recombination rates considered, whether the initial population presented an excess of coupling phase, or an excess of repulsion phase. An experimental assessment of the effect of recombination rate differences on genetic variability of quantitative traits is therefore likely to be difficult. The second study simulated a recurrent selection program in the case of high (45%), medium (19.6%) and low (9%) recombination rates. The mean genetic value increased somewhat more slowly and reached a slightly lower value for low recombination rates, whereas the genetic variance remained the highest after the second selection cycle. The evolutions in genetic mean values and genetic variances were almost identical for medium and high recombination frequencies. Thus, an allele which increases recombination rates will only be of interest in the case of close linkage or difficult crossing. Otherwise, a circular