A new technology for the specific and covalent labelling of fusion proteins with small organic molecules in vivo and in vitro has recently been developed in our laboratory. The method is based on the genetically encoded fusion of a target protein to O6-alkylguanine- DNA-alkyltransferase (hAGT). hAGT irreversibly and covalently transfers the benzyl group of O6-benzylguanine (BG) to its reactive cysteine and is subsequently degraded in vivo afterwards. hAGT possesses specific activity towards BG-based compounds which are otherwise chemically inert and have no other reaction partners in the cellular environment. This property allows for the labelling of hAGT fusion proteins with BG-fluorescein compounds inside the living cell. Mutants of hAGT that overcome several limitations of the wild-type protein such as binding to DNA, low activity, unfavourable folding properties under oxidising conditions and reaction with endogenous AGT have been generated. One mutant (named MAGT) was generated that combines most of these features with the exception of a high reactivity. In the course of this project the activity of this mutant was increased using directed molecular evolution: the phage display technology was applied to select for MAGT based highly reactive mutants. The final mutant (named SMAGT) exhibits a two-old higher activity against BG derivatives than the fastest so far described hAGT and relative to its predecessor MAGT a 16-fold higher activity. Furthermore, SMAGT combines all the described positive features of MAGT. The mutant was subsequently used for fluorescence labelling experiments on the cell surface of mammalian cells: a highly specific labelling was obtained after short incubation times with low substrate concentrations. A previously used hAGT mutant engineered only with respect to activity gave no fluorescence signal under the same conditions. SMAGT thus extends the range of possible applications of the hAGT technology. It is reported that hAGT is degraded by the ubiquitin proteasome pathway after reaction with BG substrates. Relatively few data are available concerning the post-alkylation fate of hAGT and no data at all about the in vivo stability of alkylated hAGT fusion proteins. Therefore in another part of this work the in vivo stability of hAGT fusion proteins in mammalian cells and yeast was examined: it was demonstrated that the degradation of hAGT after alkylation is strongly dependent on the C-terminal fusion partner of the protein. Furthermore it was shown that the addition of degradation signals to the hAGT (such as a N-end rule degron) can enhance the BG-induced degradation in vivo. These experiments are the first steps towards the generation of a hAGT with optimised degradation properties. Such a protein would be suitable to generate a system for the BG-induced knock-out of hAGT fusion proteins. Gene- and protein knock-out systems are powerful tools to study the physiological function of a protein in an organism. It would be highly desirable to have a system that allows the conditional knock-out of a target protein induced by an external stimulus such as a small molecule that can be used in higher organisms like mice.