Host breadth is often assumed to have no evolutionary significance in broad interactions because of the lack of cophylogenetic patterns between interacting species. Nonetheless, the breadth and suite of hosts utilized by one species may have adaptive value, particularly if it underlies a common ecological niche among hosts. Here, we present a preliminary assessment of the evolution of mycorrhizal specificity in 12 closely related orchid species (genera Goodyera and Hetaeria) using DNA-based methods. We mapped specificity onto a plant phylogeny that we estimated to infer the evolutionary history of the mycorrhiza from the plant perspective, and hypothesized that phylogeny would explain a significant portion of the variance in specificity of plants on their host fungi. Sampled plants overwhelmingly associated with genus Ceratobasidium, but also occasionally with some ascomycetes. Ancestral mycorrhizal specificity was narrow in the orchids, and broadened rarely as Goodyera speciated. Statistical tests of phylogenetic inertia suggested some support for specificity varying with increasing phylogenetic distance, though only when the phylogenetic distance between suites of fungi interacting with each plant taxon were taken into account. These patterns suggest a role for phylogenetic conservatism in maintaining suits of fungal hosts among plants. We stress the evolutionary importance of host breadth in these organisms, and suggest that even generalists are likely to be constrained evolutionarily to maintaining associations with their symbionts.