Microsatellite null alleles are common in population genetics studies, but their impact on estimating population differentiation is poorly understood. This study uses computer simulations based on the coalescent to investigate the evolutionary dynamics of null alleles, their effect on FST and genetic distances, and the efficiency of null allele frequency estimators. It also evaluates existing methods for correcting genotype data for null alleles and compares them with a new method for estimating FST. Null alleles are likely to be found in populations with large effective sizes, high mutation rates in flanking regions, and those that have diverged from the population from which the cloned allele state was drawn and the primers were designed. When populations are significantly differentiated, FST and genetic distances are overestimated in the presence of null alleles. The algorithm from Dempster et al. (1977) accurately estimates null allele frequency. The conventional method for correcting genotype data for null alleles does not provide accurate estimates of FST and genetic distances. However, using the genetic distance of Cavalli-Sforza and Edwards (1967) corrected by the conventional method gives better estimates than those without correction. FST estimation from corrected genotype frequencies performs well when restricted to visible allele sizes. Both the proposed method and the traditional correction method have been implemented in a free program available at http://www.montpellier.inra.fr/URLB/. The study used two published microsatellite data sets to confirm simulation results. The presence of null alleles was confirmed in these data sets, and their impact on population differentiation was assessed. The results show that null alleles can significantly affect the estimation of population differentiation, leading to overestimation of FST and genetic distances. The new method for estimating FST in the presence of null alleles (ENA) performs better than the conventional method (INA). The study also shows that the conventional assumption of a single null allele size common to all populations can lead to biased estimates of FST and genetic distances. The new method, which excludes null allele sizes, provides more accurate estimates. The study concludes that null alleles can significantly affect the estimation of population differentiation and that the new method for estimating FST in the presence of null alleles is more accurate.Microsatellite null alleles are common in population genetics studies, but their impact on estimating population differentiation is poorly understood. This study uses computer simulations based on the coalescent to investigate the evolutionary dynamics of null alleles, their effect on FST and genetic distances, and the efficiency of null allele frequency estimators. It also evaluates existing methods for correcting genotype data for null alleles and compares them with a new method for estimating FST. Null alleles are likely to be found in populations with large effective sizes, high mutation rates in flanking regions, and those that have diverged from the population from which the cloned allele state was drawn and the primers were designed. When populations are significantly differentiated, FST and genetic distances are overestimated in the presence of null alleles. The algorithm from Dempster et al. (1977) accurately estimates null allele frequency. The conventional method for correcting genotype data for null alleles does not provide accurate estimates of FST and genetic distances. However, using the genetic distance of Cavalli-Sforza and Edwards (1967) corrected by the conventional method gives better estimates than those without correction. FST estimation from corrected genotype frequencies performs well when restricted to visible allele sizes. Both the proposed method and the traditional correction method have been implemented in a free program available at http://www.montpellier.inra.fr/URLB/. The study used two published microsatellite data sets to confirm simulation results. The presence of null alleles was confirmed in these data sets, and their impact on population differentiation was assessed. The results show that null alleles can significantly affect the estimation of population differentiation, leading to overestimation of FST and genetic distances. The new method for estimating FST in the presence of null alleles (ENA) performs better than the conventional method (INA). The study also shows that the conventional assumption of a single null allele size common to all populations can lead to biased estimates of FST and genetic distances. The new method, which excludes null allele sizes, provides more accurate estimates. The study concludes that null alleles can significantly affect the estimation of population differentiation and that the new method for estimating FST in the presence of null alleles is more accurate.