Host-parasite checklists are essential resources in ecological parasitology, and are regularly used as sources of data in comparative studies of parasite species richness across host species, or of host specificity among parasite species. However, checklists are only useful datasets if they are relatively complete, that is, close to capturing all host–parasite associations occurring in a particular region. Here, we use three approaches to assess the completeness of 25 checklists of metazoan parasites in vertebrate hosts from various geographic regions. First, treating checklists as interaction networks between a set of parasite species and a set of host species, we identify networks with a greater connectance (proportion of realized host–parasite associations) than expected for their size. Second, assuming that the cumulative rise over time in the number of known host–parasite associations in a region tends toward an asymptote as their discovery progresses, we attempt to extrapolate the estimated total number of existing associations. Third, we test for a positive correlation between the number of published reports mentioning an association and the time since its first record, which is expected because observing and reporting host–parasite associations are frequency-dependent processes. Overall, no checklist fared well in all three tests, and only three of 25 passed two of the tests. These results suggest that most checklists, despite being useful syntheses of regional host–parasite associations, cannot be used as reliable sources of data for comparative analyses.