<p>In ecotoxicology it is common practice to apply dose-response analysis to growth inhibition data obtained from algal toxicity tests. Growth inhibition is a derived endpoint response based on growth rates that were normalized to a control. However, the normalization of growth rates introduces correlation because the resulting growth inhibition data are not mutually independent, as was the case for the original growth rate data. There has not been much work addressing this shortcoming of the standard statistical analysis. In particular, there is no literature systematically investigating the consequences of normalization on the estimation of relevant effective doses. This study systematically explored the consequences of using growth inhibition data and also explored alternative approaches. Based on both theoretical and empirical results, it was shown that it is preferable to incorporate normalization as part of the statistical analysis or, even better, entirely avoid normalization in the first place. Using a model based on growth rates allowed estimation of effective doses for growth inhibition while avoiding downwards biased estimates and too small standard errors that could both have resulted from using normalization in combination with a standard dose-response analysis.</p>