The Jones-Mollison method was applied to estimate bacterial and fungal biomass in arable soil. Also examined were: 1) applicability of the line-intercept method for estimation of hyphal length; 2) relationship between dispersion time of soil and microbial counts; 3) selection of staining dye and microscopy; 4) recovery of microorganisms added to soil; and 5) mean biovolumes. Sums of bacterial-fungal biomass estimated bimonthly for 14 months were 26–36, 36–48, 36–66 and 37–83 mg dry weight/100 g dry soil in a barley-peanuts field applied with well rotted artificial compost at a rate of 0, 20, 40 and 80 t ha−1, respectively. The biovolume-biomass conversion factor used was 0.22. Significant fluctuations of microbial biomass were observed in soil with compost. Bacterial biomass reached a maximum just after compost application in October, followed by a fungal maximum in March. Throughout the year fungal biomass was 1.7–7.1 times higher than bacterial mass. The apparent amounts of nitrogen temporarily trapped in bacterial and fungal biomass (nitrogen in maximum biomass minus that in biomass just before compost application) were calculated as 2.9 and 5.1, 4.4 and 33.5 or 9.7 and 50.9 kg ha−1in the plot applied with 20, 40 or 80 t ha−1of compost, respectively, when nitrogen content of biomass is assumed as 10% of dry matter.