We evaluate some common simulation procedures as well as a recently developed likelihood method used for testing hypotheses regarding microsatellite evolution. Results from simulated data revealed that the tests for the detection of multi-step mutations in general have some power, whereas tests for the presence of constraints on the repeat number have only very limited power. The tests were applied to population data obtained from nine different baleen whale populations. High agreement was found between results obtained using the simulation-based approach and results obtained using a likelihood ratio test. In four of the nine population samples the tests rejected the one-step mutation model. In two instances the significant deviation was due to excess of heterozygosity and in two instances to a reduced level of heterozygosity relative to the expectations under the stepwise mutation model. The former significant deviation was consistent with occasional multi-step mutations, whereas the latter may indicate the presence of constraints on the number of repeats.