We studied microbial community composition in a primary successional chronosequence on the forefront of Lyman Glacier, Washington, United States. We sampled microbial communities in soil from nonvegetated areas and under the canopies of mycorrhizal and nonmycorrhizal plants from 20- to 80-year-old zones along the successional gradient. Three independent measures of microbial biomass were used: substrate-induced respiration (SIR), phospholipid fatty acid (PLFA) analysis, and direct microscopic counts. All methods indicated that biomass increased over successional time in the nonvegetated soil. PLFA analysis indicated that the microbial biomass was greater under the plant canopies than in the nonvegetated soils; the microbial community composition was clearly different between these two types of soils. Over the successional gradient, the microbial community shifted from bacterial-dominated to fungal-dominated. Microbial respiration increased while specific activity (respiration per unit biomass) decreased in nonvegetated soils over the successional gradient. We proposed and evaluated new parameters for estimating the C use efficiency of the soil microbial community: "Max" indicates the maximal respiration rate and "Acc" the total C released from the sample after a standard amount of substrate is added. These, as well as the corresponding specific activities (calculated as Max and Acc per unit biomass), decreased sharply over the successional gradient. Our study suggests that during the early stages of succession the microbial community cannot incorporate all the added substrate into its biomass, but rapidly increases its respiration. The later-stage microbial community cannot reach as high a rate of respiration per unit biomass but remains in an "energy-saving state," accumulating C to its biomass.