Comparisons among patterns exhibited by functionally distinct genetic markers have been widely used to infer the impacts of demography and selection in structuring genetic variation in natural populations. However, such multilocus comparisons remain an indirect evaluation of selection at particular candidate loci; ideally, the identification of a candidate gene by comparative genetic methodologies should be complemented by functional analyses and experimental manipulations of genotypes in the laboratory or field. We examined genotype frequency variation among replicated intertidal habitats at two spatial scales in the grazing snail Littorina obtusata. Both of the candidate allozyme markers varied predictably with environment, and these patterns were consistent at both spatial scales. Three of four reference loci were spatially homogeneous, but one microsatellite exhibited significant structure at both geographical and mesoscales. To initiate a direct examination of whether the observed genotype frequency variation at one of the candidate markers, mannose-6-phosphate isomerase (MPI), was impacted by differential survivorship of genotypes, we conducted a series of laboratory-based thermal stress assays using snails from two geographically disparate source populations. When snails were exposed to bouts of thermal/desiccation stress, patterns of mortality were nonrandom with respect to MPI genotype. Furthermore, patterns of mortality in the laboratory manipulation coincided with the observed distribution of genotypes in the field. The data suggest the operation of selection at the Mpi or a linked locus, but functional studies and further experimentation are required to establish the relationship between MPI genotype and fitness across heterogeneous intertidal environments.