Conversion of forests of the humid tropics into pastures and agricultural fields is expected to produce, in the long-term, a decline in soil organic matter content and soil fertility. Changes in microbial biomass (Biomass C, Bc) following vegetation removal can provide an early indication for slower, less easily detectable SOM changes. Microbial biomass can also provide an index of soil fertility because it represents an important labile pool of soil nutrients and plays an active role in preventing nutrient loss. There are few published measurements of Bc in the humid tropics and fewer of Bc changes due to shifts in vegetation cover. We measured Bc in two humid tropical soils (an oxic Humitropept at 4.5% C and pH of 5.0 and a fluventic Dystrandept at 2.6% C and pH of 6.4) subjected, for 3 yr, to extreme treatments: soil maintained bare, annual harvest of re-growth, and native 20 yr old secondary vegetation. Both soils showed a similar pattern in total SOM and Bc decline following vegetation removal: after 3 yr, total C and N were reduced by 20%. Response of Bc was more pronounced. In the bare soil, most of the decline in Bc occurred within the first 6 months (to 50% of initial values) and after 15 months, Bc appeared to have stabilized at cn 35% of its initial value. Response of Bc to the annual harvest treatment was more moderate than to the bare soil treatment. Determining the precise size of microbial biomass is difficult because of variability in time and differences between techniques, but Bc values determined by two techniques and on numerous dates were high: in control plots around 2000 (oxic Humitropept) and 1300 (fluventic Dystrandept) mu g C g(-1) soil (or 250 and 187 g m(-2) after correction for differences in bulk density), which suggests that Bc in humid tropical soils can be high. Microbial C represented ca 4% of the total C in the control and declined to ca 1.5% of the total C in the bare soil. Eucaryote: procaryote ratios were close to 1 for all treatments except in the Fluventic Dystrandept control, which supported a woody vegetation and had a ratio of 3.3. This study demonstrates the dynamic nature of microbial biomass following tropical forest clearing and its potential importance for affecting nutrient loss.