The regular echinoid genera Eucidaris, Diadema and Echinometra are represented on the two sides of Central America by geminate species, believed to have resulted from the emergence of the Isthmus of Panama in the late Pliocene. Divergence between the members of each geminate pair (and of an additional Caribbean species of Echinometra from its congeners) was studied electrophoretically and morphometrically in an effort to gain an understanding of the changes in structural genes and external anatomy in populations isolated by a geographic barrier for a known period of time. Analysis of 18 presumptive loci (15 in Eucidaris), encoding a total of 13 enzymatic proteins, revealed pronounced differences in degree of differentiation in the three species pairs. Pacific populations of Diadema have diverged from their Atlantic counterparts no more than they have from populations on the same coast. Eucidaris and Echinometra, on the other hand, exhibit interoceanic genetic distances 16 and 37 times greater than intraspecific ones. Transisthmian distance in Echinometra is 20 times larger than it is in Diadema. The third species of Echinometra, E. viridis, has diverged from its sympatric Caribbean congener, E. lucunter, only one-fifth as much as the latter has diverged from its Pacific congener, E. vanbrunti. Morphometric differentiation between the members of each pair, assessed on approximately 20 characters and quantified with the Mahalanobis generalized distance, is not substantially different from local variation within each species. The contention of previous authors that morphological evidence argues for a geminate relationship of these species is, therefore, confirmed. Discriminant analysis indicates that populations of geminate species can be distinguished from each other, but that the variation which aids in this discrimination is not substantially different from local variation within each species. The ratio (but not the absolute values) of inter- to intraspecific mean Mahalanobis distances is lowest in Diadema, intermediate in Eucidaris, and highest in Echinometra, a pattern that agrees with the one displayed by the average Nei's indices calculated from electrophoretic data. This is the only instance of congruence between molecular and morphological data, and it is limited to interoceanic comparisons. The Caribbean species of Echinometra show no concordance between the two sets of characters. While on the molecular level, E. viridis has diverged little from E. lucunter, the mean morphological distance between them is twice as large as their mean intraspecific distances, a magnitude of differentiation that surpasses that of E. lucunter from the eastern Pacific E. vanbrunti. This pattern may result from different sensitivities of each level of integration to different components of the environment: allozyme frequencies may be primarily influenced by physical variables, while morphology is more likely to reflect the type of substratum that each species occupies. Rates of divergence on the two levels are, therefore, judged to be independent of each other; they only vary in unison when the components of the environment to which each is related also vary in parallel. Divergence in allopatry seems to have proceeded in rates dependent on the environmental differences as each genus (and each level of integration) has perceived them. Divergence in sympatry has been more rapid on the morphological level, possibly because of habitat separation between the closely related congeners. In a recent article Vawter et al. (1980) have claimed that data from geminate species of fish support the molecular clock hypothesis; they also criticized my suggestion (Lessios, 1979a) that the sea urchin data are inconsistent with its predictions. I present arguments as to why, in my opinion, the fish data are not adequate to test this hypothesis, while the conclusions drawn from the sea urchin data should stand.