Domesticated animals provide a unique opportunity to identify genomic targets for artificial selection to a captive environment. Nile tilapia () is a useful model species for studying the genetic basis of recent, ongoing domestication, as reared Nile tilapia strains have experienced intense artificial selection for only a short period of time. In this study, we screened 12 microsatellite loci from 11 independently reared strains (6 used for selective breeding and 5 for ranching) and their wild progenitor population of Nile tilapia to identify recent selection footprints related to domestication and selective breeding. -outlier tests were implemented using three different genetic software programs (LOSITAN, ARLEQUIN, and BAYESCAN) to identify loci under positive selection. All approaches assumed that directional selection increases genetic differentiation between populations and reduces variability at linked loci. However, because all of the tests are based on different assumptions, identifying outlier loci simultaneously using all approaches provides additional support for the candidate status of a particular locus. The LOSITAN program uses coalescent simulations to generate a neutral joint distribution of F_STheterozygosity ratios are candidates for having experienced selection. Coalescent simulations were performed us­ing 12 samples and a sample size of 30, assuming island and stepwise mutation models. The mean value was used with other values close to the mean to obtain half of the data points above and half of those below the median, as suggested in the software manual. The second method to detect selection footprints was based on the hierarchical island model and was implemented in ARLEQUIN software. Coalescent simulations were performed us­ing 12 samples and a sample size of 30, assuming a stepwise mutation model. R-project software was used to generate a neutral joint distribution of and het­erozygosity. The third method to detect selection footprints also identified loci that exhibit extreme differentiation compared with the remainder of the genome using the Bayesian likelihood method implemented via the reversible jump Markov chain Monte Carlo method. This approach was implemented in BayeScan software. was modeled using a logistic regression model and locus and population effects byrelaxing the assumption of the symmetrical island model and allowing for asymmetries in population structure. The estimates of the model parameters were automatically adjusted during short pilot runs (10 pilot runs with 5000 iterations each). We used a burn-in of 50, 000 iterations anda total chain length of 500000 iterations to identify loci under selection. BayeScan estimates the posterior probability that a locus is under selection by calculating the Bayes factor, which is given by the ratio of the posterior model probabilities of two models(selected/neutral), given the data. According to Jeffreys' interpretation, if the Bayes factor BF is > 10[lg(BF) > 1], “strong evidence” favors one model over the other and corresponds to a posterior probability > 0.91.The results showed that two loci (OMO043 and OMO114) provided significant evidence of positive artificial selection in four genetically improved NEW GIFT strain populations. Another two loci (OMO049 and OMO100) showed significant evidence of positive artificial selection in two selective breeding stocks produced by three companies. However, only one loci (OMO013) experienced positive artificial selection in five domesticated populations. The number of loci detected under selection in selective breeding populations was larger than that in domesticated populations. The loci detected under selection varied among the selective breeding and domesticated populations. These results indicate that the Nile tilapia genome has been affected by artificial selection in different directions.