In the characterization of newly identified proteins, subcellular localization studies can provide important hints to the proteins' metabolic functions. Depending on the biochemical task of an enzyme, certain subcellular environmental conditions as pH or availability of cofactors and substrates have to be fulfilled. Consequently, misdirected proteins often cannot conduct the proper chemical reaction. This study is aimed at detecting differences in bioinformatic and wet lab experiments and presenting ways for reliable analysis of subcellular localization. On a set of ten enzymes from the short-chain dehydrogenase/reductase (SDR) superfamily, we have performed predictions and experimental analyses of subcellular localization. Exemplarily, we show the localization studies on rat short-chain dehydrogenases/reductases dhrs7b and dhrs8. We demonstrate in particular that all of the prediction algorithms tested failed to assign the SDR enzymes to an experimentally verified subcellular compartment.