Two reference models are commonly used to predict mixture toxicity, Concentration Addition and Independent Action. For accurately predicting mixture effects, both reference models need a full description of the dose-response curve for all single chemicals present in the mixture. We studied the mixture effect of nickel and chlorpyrifos on survival of the soil-dwelling collembolan Folsomia candida exposed for 7 weeks in Lufa 2.2 soil. Especially chlorpyrifos toxicity showed extremely steep dose-response curves, making it impossible to fulfil the assumptions required by both reference models. Nevertheless, we showed that by monitoring the development of toxicity in time for both the single compounds and the mixtures it was possible to determine mixture effects. Even at exposure levels below the No Effect Concentration, chlorpyrifos was found to reduce nickel toxicity. In addition, this study clearly showed that toxicity is a dynamic process and that mixture effects may be dependent on exposure time. © 2010 Elsevier Ltd. All rights reserved.