Two nematode species (Cruznema tripartitum and Acrobeloides bodenheimeri) were selected to test the hypotheses that bacterial-feeding nematodes affect bacterial biomass and activity and that this feedback effect varies with nematode species and population size. For each species, nematodes of three initial population sizes were inoculated onto bacterial colonies in separate microcosms. Nematode population, bacterial biomass and CO, production were monitored in parallel microcosm settings. The responses of bacterial biomass to nematode species were different. Bacterial biomass increased significantly on d 8 in the presence of Acrobeloides when its initial numbers were 20 and 100 per microcosm; and bacterial biomass increased significantly on d 4 in the presence of Cruznema when its initial numbers were 5 and 20 per microcosm. Daily CO2 production of the microcosms with initial population sizes of 5, 20 and 100 Cruznema or of 5 and 100 Acrobeloides was significantly greater than that in microcosms without nematodes. However, the CO2 production of the microcosms with initial population of 20 Acrobeloides was not significantly different from that of the microcosms without nematodes. The increase in daily CO2 production per microcosm by Cruznema was generally greater than that by Acrobeloides for the first few days of the experiment. Nevertheless, the increase in daily CO2 production by an individual nematode was similar for both species and was a decreasing function of the initial nematode numbers. The feedback effect of each nematode species on its bacterial prey was estimated by fitting both bacterial biomass and CO2 production data to a model. Model outputs demonstrated that the feedback effect of Cruznema on bacteria was greater than that of Acrobeloides during the course of the experiment and the feedback effect of each species was not linearly correlated to initial nematode population sizes. Cruznema increased bacterial biomass and activity by a factor of 3.75-4.55 over the first 4 d, while Acrobeloides increased it by a factor between 1.97 and 3.40. (c) 2005 Published by Elsevier Ltd.