Predicted no-effect concentration (PNEC) is the key for ecological risk assessment and has become a focus of the study. Generally, extrapolation was used to derive PNEC values by different evaluation approaches according to results of single species toxicity test. However, PNEC values were different for the same toxicity data for the same chemicals if extrapolation methods were different. To evaluate PNEC values derived by different assessment methods, PNEC values of pentachlorophenol (PCP) were calculated using three different approaches, i. e. point estimation, statistical evaluation as well as interval evaluation. At the same time, the main factors controlling PNEC values were analyzed. The results showed that PNEC values for PCP by different assessment factors were 0.420 microg x L(-1) for acute toxicity data and 0.500 microg x L(-1) for chronic toxicity data, respectively. The PNEC values for PCP by statistical evaluation were 10.1 microg x L(-1) for probit unit and 8.10 microg x L(-1) for logit unit. The values derived by interval evaluation were 6.00 microg x L(-1) for PNEC-L95% and 11.8 microg x L(-1) for PNEC-L50%, respectively. PNECs obtained by the point estimation were the lowest and lower one order of magnitude than the PNEC values obtained by the other two methods. The PNEC values derived by the statistical assessment and interval evaluation were very similar. In addition, toxicity data type and statistical models have little effect on PNEC value for PCP with a great body of toxicity data. The results indicated that statistical evaluation and interval evaluation methods should be selected to derive PNEC values for well-studied substances like PCP, while point estimation should be used in preliminary screening assessment of chemicals to determine whether the effect is of concern.