One arable and two grassland soils were sieved < 2 mm, air-dried and subsamples rewetted to a range of water contents. They were then incubated for 7 days at 25°C before mycelial (M, fungal and actinomycete organisms) and bacterial (B, discrete organisms < 1.5 p.m 3 ) volumes were estimated by light microscopy. Mycelial and bacterial volumes, surface areas, and masses, and their respective ratios all differed significantly between soils, but not between water contents within soils. Bacterial volume (mean 0.10, modal 0.07 p.m 3 per bacterium) and morphology (890/0 of bacteria were coccal) were very similar for all three soils. However, mycelial populations differed between soils. Modal mycelial diameter was 1.5 p.m for all soils, but mean diameter was 2.8 p.m for one soil and 3.8 p.m for the other two soils. Total surface area was dominated by bacteria, whereas volume and mass were dominated by mycelia. In the two soils with similar microbial populations, total (i.e., visible) microbial volume (T, i.e., M + B) and mycelial:bacterial volume ratios (R, i.e., MIB) were related: are corrected for the apparently dead mycelia then R (= 0.67), the ratio of viable mycelial:bacterial volume, is independent of T, the total microbial volume. Inability to discriminate non-viable volume may have obscured significant correlations between water content and the microbial populations of these soils.