We used molecular techniques to examine the ectomycorrhizal fungal community associated with pinyon pine (Pinus edulis) growing in two soil types in a semiarid region of northern Arizona: cinder soils low in nutrients and moisture, and sandy-loam soils with higher moisture and nutrient levels. Pinyon performance (e.g., growth, reproduction, water stress) has been shown to be markedly lower in cinder than in sandy-loam environments. Fungal community composition and richness were determined using RFLP (restriction fragment length polymorphism) analysis of ectomycorrhizal root tips collected from three sites within each soil type. Several patterns emerged from these analyses. First, communities in both cinder and sandy-loam soils were dominated by one or a few abundant ectomycorrhizal types, a species abundance pattern common to many plant and animal communities. Second, unlike the pattern for many other organisms, ectomycorrhizal fungal type (species) richness was not correlated with measures of ecosystem productivity such as soil nutrient and moisture levels; cinder and sandy-loam soils had similar numbers of ectomycorrhizal fungal types (range of 15–19 fungal types for both soil types). Third, soil type and fungal community composition were linked, as cluster analysis demonstrated greater similarity of fungal communities from sites within soil types than between them. Differential amplification using primers with enhanced specificity for basidiomycetes indicated that an average of 85% of the ectomycorrhiza found at the sandy-loam sites were members of the subphylum Basidiomycotina, whereas over half (mean = 52%) of the ectomycorrhiza at the cinder sites were formed by members of other fungal groups, probably the subphylum Ascomycotina. Fourth, a preliminary survey of 14–45 ectomycorrhizal root tips from each of 20 trees at one cinder site indicated that trees were dominated by one or a few ectomycorrhizal RFLP types. However, the same RFLP types did not dominate on all trees, and dominant types showed considerable spatial variation. Fifth, the RFLP patterns of some fungal sporocarps matched those of ectomycorrhizal root tips, but many did not, indicating that many of the ectomycorrhizal fungi at these sites fruit infrequently, whereas other fungi with more abundant sporocarps may not form ectomycorrhiza. This emphasizes the need to characterize the ectomycorrhizal communities formed on the plant roots themselves, rather than characterization based on sporocarps alone, particularly in arid environments. Finally, the differences in ectomycorrhizal fungal communities we observed between soil types supported the concept that conserving fungal diversity requires conservation of host plant species over their entire range, not just typical sites. If future studies corroborate these patterns, our results suggest that abiotically stressful environments are important to include in these conservation efforts.