1. In order to find a more reliable basis for ecotoxicological risk assessment, two microcosm experiments, one in the laboratory and another in the held (lysimeters), were performed using pentachlorophenol (PCP) as a model contaminant. PCP was applied at three (0, 50 and 500 mu g g(-1) dry soil in the laboratory experiment) or at two (0 and 50 mu g g(-1) in the lysimeters) concentrations in raw humus forest soil and its effects on soil organisms and decomposition processes were monitored. Soil fauna was manipulated at two levels in the laboratory experiment: simple and diverse communities. 2. PCP was strongly adsorbed onto the humus. 3. Microbes, nematodes, enchytraeids and predatory gamasid mites were sensitive to PCP in the laboratory experiment. 4. No significant faunal effects were observed on carbon and nutrient mineralization. At the highest PCP concentration carbon liberation (CO2-evolution) was lowered and NH4+-N was accumulated in the soil. PCP contamination indirectly affected soil pH and water content of the organic soil layer. 5. In the lysimeters, after 15 weeks there were fewer enchytraeids and prostigmatid mites in the soil contaminated with PCP, but no changes in nitrogen mobilization were observed. After 49 weeks, however, there was less NH4+-N in the contaminated soil, while no differences in faunal composition were found. 6. The results showed that, in comparison with single-species toxicity tests, more ecologically relevant microcosm experiments give important information about the behaviour of the decomposition system under contamination. This information, including also indirect impacts of chemicals, may be useful for the improvement of ecotoxicological risk assessment.