Evolutionary and conservation biologists often use molecular markers to evaluate whether populations have experienced demographic bottlenecks that resulted in a loss of genetic variation. We evaluated the utility of microsatellites for detection of recent, severe bottlenecks and compared the amounts of genetic diversity lost in bottlenecks of different sizes. In experimental mesocosms, we established replicate populations by releasing 1, 2, 4 or 8 pairs of the western mosquitofish, Gambusia affinis (Poeciliidae). Using eight polymorphic microsatellite loci, we quantified seven indices of genetic diversity or change that have been used to assess the effects of demographic bottlenecks on populations. We compared indices for the experimentally bottlenecked populations to those for the source population and examined differences between populations established with different numbers of founders. Direct count heterozygosity and the proportion of polymorphic loci were not very sensitive to genetic changes that resulted from the experimental bottlenecks. Heterozygosity excess and expected heterozygosity were useful to varying degrees in the detection of bottlenecks. Allelic diversity and temporal variance in allele frequencies were most sensitive to genetic changes that resulted from the bottlenecks, and the temporal variance method was slightly more correlated with bottleneck size than was allelic diversity. Based on comparisons to a previous study with allozymes, heterozygosity, temporal variance in allele frequencies and allelic diversity, but not proportion of polymorphic loci, appear to be more sensitive to demographic bottlenecks when quantified using microsatellites. We found that analysis of eight highly polymorphic loci was sufficient to detect a recent demographic bottleneck and to obtain an estimate of the magnitude of bottleneck severity.