The final glycosylation pattern of recombinant proteins might be determined by the choice of the host cell, cell culture conditions and the purification. Each step influences glycosylation and improved glycosylation of the product may thus be obtained by changing relevant parameters of the respective step. In the present study the glycosylation of two pharmaceutically relevant glycoproteins was studied with respect to host cell, culture conditions and purification. The limiting step in the process was identified. The development of a technique was initiated for the high-throughput screening of cells showing favourable glycosylation characteristics. Erythropoietin (EPO) and immunoglobuline G1 (IgG1) were produced using different cell-types and CHO clones. According to high-pH anion exchange chromatography (HPAEC) the IgG1 produced by different CHO clones generally carried complex-type biantennary N-glycans with a clone-dependent degree of terminal galactosylation. Core fucosylation was essentially complete. According to HPAEC, HEK-293 cells produced a qualitatively identical pattern, albeit with a higher level of galactosylation and possibly a small amount of afucosyl-N-glycans or N-glycans carrying GlcNAci. N-glycans derived from IgG1 produced in SP2/0 cells were extended additionally with terminal α-galactose and terminal NeuAc. N-glycans from EPO produced by different CHO clones consisted generally of bi-, tri-, and tetraantennary N-glycans with a variable degree of terminal sialylation. Core fucosylation was essentially complete. The results indicated clearly a cell-line and cell-clone dependent glycosylation. Both glycoproteins were produced by CHO cells under different culture conditions. For a variety of parameters the glycosylation remained unchanged. For IgG1 and EPO the temperature was identified as one parameter influencing the oligosaccharide structures. Using EPO as model glycoprotein intracellular nucleotide-sugar availability and its effects on glycosylation were tested. Due to low intracellular concentrations of nucleotide-sugars these could not be measured directly but precursor feeding experiments, whereby levels of intracellular UDP-N-acetylhexosamine and CMP-N-acetylmannosamine could be increased, suggested that the changes in glycosylation at different temperatures can not be explained by changed intracelullar nucleotide-sugar availability only. The glycosylation of IgG remained constant during the purification process. For EPO it was shown that sialylation may be improved by employing anion-exchange chromatography and hydroxyapatite chromatography. A major drawback of this approach were the low recoveries of EPO. In summary the results indicated that adequate glycosylation of recombinant proteins is best addressed at the choice of the host cell. The development of a screening technique was started allowing the screening of cells producing the protein with a favourable glycosylation pattern. Single cells may be immobilised in alginate beads. On the surface of these beads specific antibodies directed against the product are located. Captured product may be analyzed using carbohydrate-specific lectins and screening may be automated using a FACS. Because of time limitations this work could not be finished. Using secretory component as model glycoprotein it was shown however that the principle of the detection has potential for further development.