Critical analysis of methods for assessment of predicted no-effect concentration.

Current risk assessment procedures for chemical substances are based on hazard quotients and require determination of a predicted no-effect concentration (PNEC). These concentrations are usually computed by applying an assessment factor to the lowest available toxicity value. However, other approaches have also been proposed based on statistical distribution of ecotoxicity data. This paper compares the various approaches in terms of precision and robustness. When only a few data are available, an assessment factor approach can be used. However, whenever possible (i.e. for a large set of chronic data), a statistical approach like that developed by T. Aldenberg and W. Slob (1991, Confidence Limits for Hazardous Concentrations Based on Logistically Distributed NOEC Toxicity Data, RIVM Report 71902002) with a 50% confidence level is the most precise and provides a stable value with increasing confidence when the number of tests increases.