The development of vesicular arbuscular (VA) mycorrhizas was followed for subterranean clover grown in 20 field soils in a glasshouse experiment. The aims of the study were: to understand the way in which mycorrhizas develop in field soils; to identify those factors which could be used to predict field sites suitable for inoculation with VA mycorrhizal fungi. In each soil, the amount and rate of mycorrhiza formation were estimated for species of each genus represented. The data were examined in relation to differences among soils in numbers of spores of VA mycorrhizal fungi and in soil properties. A poor correlation between total spore numbers and the total amount and rate of infection formed was attributed to two factors. First, infection by fine endophyte (a fungus which does not form large spores that can be counted) was ubiquitous, but the amount of mycorrhizas formed by this fungus varied greatly. Second, species of fungi differed in their rates of infection. In general, there was an association between spore numbers and infection development for individual fungal species. The development of mycorrhizas in any soil fell into one of three categories: I, rapid and extensive; II, extensive but with a lag phase; III, slow and limited in extent. The species of fungi in soils from each category were similar. Category I included the soils which were most deficient in phosphorus for plant growth. However, from measurements of soil properties alone, it is not possible to predict those soils which are suitable for the introduction of inoculant VA mycorrhizal fungi.