The suitability of the fumigation-incubation (FI), fumigation-extraction (FE) (with specific k-factor determined by C-14-labelling) and substrate induced respiration (SIR) methods to obtain measures of the microbial biomass in forest soils from the Amazonian basin was examined. Several representitive topsoil (0-10 cm) samples were studied. A reasonable estimate of microbial biomass C in two acid Oxisols (pH <4.5) with a high clay content was obtained by the FI method, provided no correction was applied for basal CO2 respiration. However, on two Ultisols and one Alfisol more consistent estimates of microbial C were obtained by the FI method, where correction for basal respiration was obtained using the respiration of non-fumigated (NF) soils between 0-10 d, or fumigated (F) soils between 10-20 d. Estimates of microbial C by the FE method did not differ significantly from those obtained by the FI method when calculated using a specifically-determined k-factor of 0.26 to convert the extractable-C flush to microbial biomass C, In the Ultisols and the Alfisol, the SIR method gave similar estimates of microbial C to those obtained by the FI and FE methods, but in both Oxisols microbial C was underestimated when a standard conversion factor was applied. Possibly the SIR assay was adversely affected by a high proportion of dormant organisms and a slow gaseous release of respired CO2 from these heavy clay soils. When estimated using the FE method, microbial C was in the range 890-1100 mu g C g(-1) soil in the Oxisols, 340-470 mu g C g(-1) soil in the Ultisols and around 580 mu g C g(-1) soil in the Alfisol. Microbial C comprised 3-4% of the total soil organic C and microbial N 4-7% of total N. Microbial C-to-N ratios averaged 6.9 +/- 0.8. Although the FI method can be used on these Amazonian soils, the FE method is more convenient and generally appears suitable for estimating both microbial C and N in the same extract. The SIR method is not recommened for use with Oxisols.