Microsatellites have proved to be useful for the detection of weak population structure in marine fishes and other species characterized by large populations and high gene flow. None the less, uncertainty remains about the net effects of the particular mutational properties of these markers, and the wide range of locus polymorphism they exhibit, on estimates of differentiation. We examined the effect of varying microsatellite polymorphism on the magnitude of observed differentiation in a population survey of walleye pollock, Theragra chalcogramma. Genetic differentiation at 14 microsatellite loci among six putative populations from across the North Pacific Ocean and Bering Sea was weak but significant on large geographical scales and conformed to an isolation-by-distance pattern. A negative relationship was found between locus variability and the magnitude of estimated population subdivision. Estimates of FST declined with locus polymorphism, resulting in diminished power to discriminate among samples, and we attribute this loss to the effects of size homoplasy. This empirical result suggests that mutation rates of some microsatellite loci are sufficiently high to limit resolution of weak genetic structure typical of many marine fishes.