The contribution from increased lengths of saprophytic and mycorrhizal hyphae to the aggregation of a sandy soil was assessed in a glasshouse experiment. Increased length of mycorrhizal hyphae was encouraged by inoculation of Lolium rigidum with Scutellospora calospora and the effects of the roots on aggregation were restricted by containment within 38 mu m mesh bags. The hyphae of saprophytic fungi were increased by the addition of glucose or straw substrates at 2.3 mg g(-1). After incubation of the soils for five weeks, changes in > 2 mm water-stable aggregates were assessed in the root-free soil before and after dry-sieving through 8 mm mesh, to test the resistance of the aggregates to soil abrasion. Hyphal lengths and hot-water extractable carbohydrate C contents were measured in the > 2 mm water-stable fraction and non-aggregate ( <2 mm) soil of the soils that were not dry-sieved.The treatments increased hyphal lengths from 3.2 m g(-1) to 4-6.2 m g(-1), but the hyphal lengths were not different between the treated soils (P <0.05) because much of the hyphal growth occurred only in the soil that was aggregated. The lengths of hyphae in the > 2 mm aggregates from the soils amended with straw and glucose were 2-5.7 m g(-1) greater (P <0.05) than the lengths in non-aggregate soil (4-6 m g(-1)). Inoculation of non-amended soil increased the proportion of mycorrhizal roots by 340% (P <0.05) but had little effect on hyphal length in the non-amended soils. Water-stable aggregation was increased from 7-16 g kg(-1) in the non-amended soils to 84-143 g kg(-1) in the straw- and glucose-amended soils (P <0.05). Under a Scanning Electron Microscope, sand grains in the aggregates appeared to be linked together only by hyphae. In addition, the hot-water-extractable carbohydrate C contents of water-stable aggregates and non-aggregate soil were not different, indicating little involvement of microbial polysaccharides in stabilising the aggregates, Despite the strong evidence for hyphal involvement in stabilising aggregates, only the hyphal lengths in the whole soil of the non-amended soil and > 2 mm aggregate fraction of the soil amended with glucose were significantly correlated with aggregation (P <0.05). The amounts of dry-stable aggregates in 8 mm sieved soil were generally increased by 63-147% in the amended soils compared to non-amended soils and 27 to 33% of these aggregates were water-stable aggregates indicating that hyphae also contributed to the stabilisation of dry aggregates. Increases in aggregate stability could be largely attributed to the increased growth of hyphae in the sandy soil. Hyphae were able to contribute greatly to the stability of the sand aggregates because much of the soil was > 250 mu m sand which favoured crosslinking of the sand grains by short lengths of hyphae.