Statistical methods have been proposed to detect recent bottlenecks on the basis of genetic characterizations of natural populations. In the absence of direct estimates of contemporary or historical population numbers, we tested the indirect M-ratio method based on microsatellite motif size frequency profiles using three historical and three contemporary Great Lakes populations of Salvelinus namaysuch for which severe reductions in population numbers are documented. Simulations employing plausible ranges of empirical population parameter values were used to explore bottleneck likelihood surfaces. We show that single values of the M-ratio are not sufficient to unambiguously infer a bottleneck without knowledge of mutation rates and effective population size (i.e., 4Ne mu or [symbol: see text]). Inferences of the degree of population bottleneck would be best supported if analyses were conducted across plausible ranges of [symbol: see text] and by qualitative comparisons among population samples.