Replication of the genome is strongly inhibited when high fidelity DNA polymerases encounter unrepaired DNA lesions, which can not be processed. The highly stringent active sites of these polymerases are unable to accommodate damaged bases and therefore DNA lesions block the replication fork progression. In order to overcome this problem, cells have evolved mechanisms for either repairing the damage, or synthesising past it with specially adapted polymerasases. Eukaryotic DNA polymerase eta (Pol eta), belonging to the Y-family of DNA polymerases, is outstanding in its ability to replicate through a variety of highly distorting DNA lesions such as cyclobutane pyrimidine dimers (CPDs), which are the main UV-induced lesions. Also cisplatin induced 1,2-d(GpG) adducts (Pt-GGs), which are formed in a typical cancer therapy with cisplatin can be processed by Pol eta. The bypass of such intrastrand crosslinks by high fidelity DNA polymerases is particularly difficult because two adjacent coding bases are simultaneously damaged. Thus, replication by Pol eta allows organisms to survive exposure to sunlight or, in the case of cisplatin, gives rise to resistances against cisplatin treatment. Mutations in the human POLH gene, encoding Pol eta, causes the variant form of xeroderma pigmentosum (XP V), characterized by the failure to copy through CPDs. This leads to strongly increased UV sensitivity and skin cancer predisposition. This thesis describes mechanistic investigations of the translesion synthesis (TLS) process by S. cerevisiae DNA Pol eta at atomic resolution, which were undertaken in collaboration with the Hopfner group. To study this process, cisplatin lesioned DNA had to be prepared first. Once this technique was established, the catalytic fragment of Pol eta was crystallized as ternary complex with incoming 2',3'-dideoxycytidine 5'-triphosphate (ddCTP) and an primer - template DNA containing a site specific Pt-GG adduct. The first obtained structure shows the ddCTP positioned in a loosely bound conformation in the active site, hydrogen bonded to the templating base. Realizing the importance of the 3’ hydroxy group for positioning the NTP and the DNA correctly inside the polymerase, the complex was crystallized again with a 2’-deoxynucleoside 5’-triphosphate (dNTP). To prevent nucleotidyl transfer, primer strands which terminate at the 3’-end with a 2’,3’ dideoxy ribose were prepared by reverse DNA synthesis and used for cocrystallization. The resulting crystals diffracted typically to 3.1-3.3Å resolution at a synchrotron light source. A Pol eta specific arginine (Arg73 in yeast Pol eta) was identified for its importance to position the dNTP correctly in the active site and was shown to be necessary for lesion bypass. In contrast to the fixed preorientation of the dNTP in the active site, the damaged DNA is bound flexibly in a rather open DNA binding cleft. Nucleotidyl transfer requires a revolving of the DNA, energetically driven by hydrogen bonding of the templating base to the dNTP. For the 3’dG of the Pt-GG, this step is accomplished by bona fide Watson-Crick base pairs to dCTP and is biochemically efficient and accurate. In contrast, bypass of the 5’dG of the Pt-GG is less efficient and promiscuous for dCTP and dATP. Structurally, this can be attributed to misalignment of the templating 5’dG due to the rigid Pt crosslink. In cooperation with the Cramer group the structural reasons for the blockage of RNA Polymerase II (RNAP II) by the cisplatin lesion were elucidated. Using structural as well as biochemical methods it could be shown that stalling results from a translocation barrier that prevents delivery of the lesion to the active site. AMP misincorporation occurs at the barrier and also at an abasic site, suggesting that it arises from nontemplated synthesis according to an 'A-rule' known for DNA polymerases. RNAP II can bypass a cisplatin lesion that is artificially placed beyond the translocation barrier, even in