This chapter describes the prediction of toxic effects with data mining techniques in a stepwise approach. Methods are characterized in terms of principle advantages and shortcomings and interpretability of the results. We seek to present techniques that are effective as well as universally applicable. We also give some software recommendations focusing on open-source software, which is not only free, but also transparent and extensible. All packages for the R environment for statistical computing (as well as R itself) are available from CRAN, the central R repository [1] [http://www.r-project.org].