Chemical toxicity was estimated by integrating in vitro study results with physiologically-based biokinetic models for eight neurotoxic compounds (benzene, toluene, lindane, acrylamide, parathion/oxon, caffeine, diazepam and phenytoin). In vitro studies on general and specific neurotoxicity were performed and biotransformation and tissue-blood distribution studies were used in modelling the biokinetic behaviour of the compounds. Subsequently, neurotoxicity was estimated from the integrated in vitro and kinetic studies. These results were compared with in vivo data from the literature on minimal neurotoxicity for these compounds, such as lowest-observed-effect levels (LOELs). The discrepancy between estimated and experimental LOELs ranged from 2- to 10-fold. LOEL estimates for compounds with a relatively low toxicity were more accurate than for compounds with a relatively high toxicity. LOELs for the most active compounds could only be established after consideration of additional in vitro results from the literature. The present study has generated encouraging results on the risk assessment of chemicals from in vitro studies and computer simulations and has identified some key directions for future research.