Although genetic differences throughout the geographic range of a species presumably arise from isolation and differential selection in different areas, significant differences have been observed among relatively restricted geographic areas in prolific marine organisms with planktonic dispersal stages. Given selection of sufficient intensity, genetic differences might be produced among the sedentary adult stocks even though the dispersal power of the species is such that all adult stocks collectively constitute the same panmictic population. The relative importance of selection and isolation can be estimated in the American eel, Anguilla rostrata, since all breeding occurs in a circumscribed region northeast of the West Indies and ocean currents disperse the larvae to American shores from northern South America to the Arctic. As a consequence, any genetic divergence among localities must necessarily arise in the absence of isolation. Samples of elvers and more advanced young and adults were obtained from a range of 20⚬ latitude from Vero Beach, Florida to St. John's, Newfoundland. Five loci, synthesizing the following isoenzymes, were used to estimate genetic heterogeneity among samples: alcohol dehydrogenase, sorbitol dehydrogenase, malate dehydrogenase, non-specific esterase, and phos-phohexose isomerase. Polymorphic variation at the esterase locus was the most complex and least interpretable. There was significant heterogeneity among localities of both adults and elvers. While the absolute magnitude of allele frequency differences among localities at the esterase locus was about the same as that observed at other loci, these differences did not occur in any interpretable pattern. Malate dehydrogenase is essentially homogeneous over the latitudinal range of our samples though significant heterogeneity exists in allele frequency as a function of a single heavily weighted deviant sample from Halifax, Nova Scotia. The most common allele at each of the remaining three loci vary among localities in a significant linear pattern in relation to latitude. The existence of significant differences among localities, that must arise anew among the subgroups of each age cohort as they disperse from their place of origin, demonstrates that selection pressures are sufficient to cause gene frequency changes of the order of 10% per generation. Moreover, if we follow the assumptions commonly made in deducing magnitudes of genetic variability in natural populations, we must conclude that these selective differentials characterize a large number of gene loci in this species. By implication, there must be a great recombinational load in the eel and a high variance in fitness. In addition to the relevance that these findings have to contemporary considerations of magnitudes of natural selection occurring within natural populations, they indicate that the recognition of separate Mendelian populations on the basis of significantly different allele frequencies in a number of other commercially important species, is highly questionable. It is logically possible, that the genetic differences we have observed are not due to differential survival of various genotypes, but due to some genetically based behavioral mechanism which allows pelagic larvae to cue on particular environmental signals and thereby arrive at the American coast in a non-random manner. Although none is decisive, the following considerations suggest this as an unlikely explanation for the observed genetic differences. There is considerable variation in such environmental factors as temperature, light, flow patterns and flow rates, and it there-fore appears extremely unlikely that such factors could adequately serve as cues to navigating larvae. Additional reasons for regarding any local-population hypothesis as unlikely would include a north-south gradient in sex ratio and the absence of any geographic variation in meristic and other morphological features.