the framework of the EDIT (Evaluation guided Development of In vitro Toxicity and toxicokinetic tests) programme, the long-term cytotoxicity of 27 chemicals was investigated on Hep G9 cells. The first step in the experiments was to determine the PI50(24h) of the chemicals. This is the concentration of compound needed to reduce the total protein content by 50% after 74 h of treatment. In the long-term experiments the chemicals were tested in six different concentrations, using the PI50(24h) as maximum concentration. The cells were treated twice a week with the same concentration of test compound and were trypsinised and counted once a week (dynamic culture). The number of cells was compared to the number of cells of the control. Three major long-term cytotoxicity patterns could be distinguished. After 6 weeks, the EC50(6w)s were determined. This is the concentration of compound needed to reduce the number of cells by 50% after 6 weeks of treatment. These values were compared with the PI50(24h). A good correlation was found for the 27 chemicals (r(2) = 0.860). After 6 weeks, the concentration of test compound needed to reduce the total cell protein content by 50% after 24 h after 6 weeks of pretreatment of the cells with a particular concentration of test compound was measured: the PI50(24h-6w). For the majority of compounds then is no difference between the P150(24h) and the PI50(24h-6w). For ethanol, arsenic (III) oxide, verapamil hydrochloride and orphenadrine, the PI50(24h-6w) increased in comparison to the PI50(24h). For some compounds a dose-response was observed, indicating that the cells have become more resistant or more sensitive. Linear regression analysis revealed a good correlation (r(2) = 0.709) between the EC50(6w) and the human acute toxicity. All these data indicate that a good alternative test may be found for predicting the long-term human toxicity. (C) 2001 Elsevier Science Ltd. All rights reserved.