A methodology is proposed to predict single-species toxicity in natural waters by using laboratory bioassay data which relate effect to a tissue concentration of toxicant or to injury accumulation. Such relationships should be independent of test conditions and therefore transferable from lab to field. A mathematical model of uptake and elimination of toxicants by fish is used to relate exposure concentration to tissue concentration and then to affect. Prediction of toxicity in a laboratory test in which the exposure concentration varies in time is presented as a first step in testing the methodology. Survival time under time-varying exposure is calculated and compared to observed data. The model predicted survival times that were consistent with the observed times. Whole-body residue is used as the estimate of dose.