Toxicity endpoints that are measured in natural waters with dif-ferent DOC, pH, and hardness will vary considerably. For example, EC10 values for D. magna showed an 8-fold difference among different natural waters. For the BLMs to work, they must explain these differences. The chronic Ni BLMs are able to predict toxicity among different waters within a factor less than 2, indicating that the models are well calibrated and are able to accurately predict chronic toxicity in natural freshwaters. Figure 2 shows an overview of the relationship between observed and BLM predicted chronic toxicity values for the D. magna, C. dubia, P. subcapitata, and O. mykiss BLMs.