Aims and Background The aim was to investigate the diversity and distribution of Glomeromycotan fungi forming arbuscular mycorrhizal associations (AMF) in undisturbed and disturbed habitats in the vicinity of Kakadu National Park in tropical Australia. This is a tropical region with a 7–9 month dry season and a monsoonal wet season. Complimentary methods of fungus detection were used to investigate the diversity and relative dominance of AMF at a regional scale. Methods Soils were sampled from 32 sites, representing eucalypt savanna woodlands, wetlands, sandstone escarpment, rainforest, and disturbed mine waste rock mps (overburden or spoil). Populations of AMF were identified and quantified using spores from soil. Morphology patterns of fungi colonising bait plant roots were examined and isolates were obtained by four complimentary pot-culturing methods. Results Different methods of detecting fungi produced different answers about which AMF were most important in the tested soils. In particular, spore surveys apparently underestimated the importance of Glomus species and overestimated the activity of Acaulospora species with numerous small spores, while calculated spore biovolumes overestimated the importance of Scutellospora and Gigaspora species with large spores, relative to inoculum levels of these fungus categories measured in bioassays. Spore surveys revealed 15 species of fungi and 8 additional fungi were recovered from the same soil samples using potculture isolation methods. Pot-cultures were especially important for detecting Glomus species that had high inoculum levels, but rarely produced spores in soils. Spores of AMF increased in abundance as vegetation developed in mine habitats reaching a peak that was higher than in undisturbed plant communities. Spore numbers (but not biovolumes) were well correlated with bioassay measurements of inoculum levels. Conclusions Most AMF species were widespread, but several were restricted to disturbed habitats or wetland soils. Undisturbed sites had a substantially higher diversity of AMF than partially vegetated mine waste rock dumps. It is recommended that AMF population surveys should not be based entirely on spore occurrence data, to avoid overlooking important fungi that sporulate infrequently. These fungi could be detected by bioassays or pot culture isolation from soil. Major variations in the detectability of AMF correspond to different life history strategies and can mask variations in their abundance.