We used null model analysis to test for nonrandomness in the structure of metazoan ectoparasite communities of 45 species of marine fish. Host species consistently supported fewer parasite species combinations than expected by chance, even in analyses that incorporated empty sites. However, for most analyses, the null hypothesis was not rejected, and co-occurrence patterns could not be distinguished from those that might arise by random colonization and extinction. We compared our results to analyses of presence–absence matrices for vertebrate taxa, and found support for the hypothesis that there is an ecological continuum of community organization. Presence–absence matrices for small-bodied taxa with low vagility and/or small populations (marine ectoparasites, herps) were mostly random, whereas presence–absence matrices for large-bodied taxa with high vagility and/or large populations (birds, mammals) were highly structured. Metazoan ectoparasites of marine fishes fall near the low end of this continuum, with little evidence for nonrandom species co-occurrence patterns.