The amount of Gaeumannomyces graminis var. tritici inoculum was quantified by slot-blot hybridization assay, using a specific DNA probe, pG158. Estimates of the amount of G. graminis var. tritici DNA in naturally infested and inoculated soils were made and correlated with take-all disease severity as determined by a conventional soil bioassay (expressed as a percentage of seminal roots with G. graminis var. tritici lesions). Disease severity in the glasshouse assay and DNA levels were highly correlated. The level of take-all disease severity in naturally infested soil can be predicted using the equation yi=53·078 log (xi)+88·866, where yi=the predicted value of the disease severity (%) and xi=amount of G. graminis var. tritici DNA (ng) in 100 g soil. The DNA-based assay for quantifying the amount of G. graminis var. tritici has been further developed and improved in terms of the sampling techniques and efficiency. Soils of different types were collected to a depth of 10 cm and organic matter between 0·5–1·4 mm in size was separated from this soil. Using a slot-blot hybridization assay, it was found that G. graminis var. tritici was mostly present in the organic matter fraction greater than a size of 0·5 mm and in soil at a depth of 5 cm. This rapid and reliable DNA-based assay is being used in conjunction with a model of take-all disease development to predict disease levels in the field due to G. graminis var. tritici.