Regio- and stereoselective hydroxylation of steroid molecules remains a challenge in industrial steroid hormone synthesis. Here, the use of cytochrome P450 monooxygenases is of high interest due to their remarkable ability to catalyze the direct hydroxylation of non-activated carbon atoms in a regio- and stereoselective manner. This allows also the selective oxyfunctionalization of steroids, thus avoiding the use of protecting groups and several time-consuming chemical steps. In this thesis, the use of bacterial P450s, especially CYP154C5 from Nocardia farcinica, was investigated for the selective hydroxylation of various steroids. In particular, an efficient whole-cell biocatalyst was developed based on the recombinant expression of CYP154C5 in Escherichia coli together with putidaredoxin and putidaredoxin reductase from Pseudomonas putida for electron transfer. Cofactor regeneration was achieved by the simple addition of glucose and with the help of hydroxypropyl-beta-cyclodextrin for substrate solubilization, several steroid molecules could be successfully converted with up to 15 mM initial steroid concentration using this whole-cell system. Additionally, 16alpha-hydroxylated steroids, which are important precursors for the synthesis of highly potent glucocorticoids, were selectively produced on preparative scale with total turnover numbers (TTN, µmol substrate consumed µmol-1 CYP154C5) exceeding 2000 and space-time yields of several grams per liter a day. Furthermore, CYP154C5´s crystal structure with six different steroid substrates bound in the active site was determined to identify key residues in the active site that are involved in substrate binding and therefore responsible for the remarkable selectivity of the P450 monooxygenase. Among the 21 residues forming the active site, four were suspected to play an important role in substrate binding and catalytic activity. In particular, residues M84 and F92 were identified to be involved in hydrophobic interactions with the steroid core, whereas residues Q239 and Q398 were found to form hydrogen bonds with oxyfunctional groups at positions C3 and C17 of the steroid substrate. Two strategies were applied to further investigate the selectivity of CYP154C5. Firstly, the four mentioned residues were exchanged by alanine through site-directed mutagenesis in order to remove the mentioned enzyme-substrate interactions. The resulting mutants were analyzed in the conversion of six steroid substrates by determining dissociation constants (KD), turnover numbers (TON), TTN and coupling efficiencies using purified proteins. Results confirmed the importance of the four residues for substrate binding and conversions as indicated by the decreased substrate affinity and overall efficiency of the conversion. As the only exception, nandrolone was found to be better converted by mutant Q239A as compared to wild-type CYP154C5. Interestingly, with a single mutation in the active site a secondary hydroxylation product was obtained in the conversion of progesterone by CYP154C5-F92A, thus changing the regioselectivity of this enzyme. In a second approach, CYP154C5 wild type was tested in the conversion of selected steroids lacking key functional groups in their structure that could substantially affect substrate binding and therefore also selectivity of the enzyme. Interestingly, here the regioselectivity of CYP154C5 was altered when a steroid lacking a functional group at position C17 was used as substrate as indicated by the formation of the 15alpha- instead of 16-hydroxylated product. Furthermore, the use of a steroid substrate bearing no functional group at C3 also resulted in a changed regioselectivity of CYP154C5 as here four different hydroxylation products were obtained. In contrast, steroid substrates with large substituents at position C17 were not converted by the P450. These findings shed light on the possibility to create a CYP154C5-based tool box for selective steroid hydroxylation by protein engineering. Interestingly, a much smaller substrate, beta-ionone, was also selectively monohydroxylated by this enzyme. CYPs from other bacterial sources were also tested in steroid conversions within this thesis. As novelty, we could show for the first time that also 3beta-hydroxy-Delta5-steroids are hydroxylated by CYP106A2 from Bacillus megaterium in a regioselective manner. Furthermore, CYP110D1 from Nostoc sp. was shown for the first time to hydroxylate also other steroid molecules than testosterone. In summary, industrially important target positions for steroid hydroxylation such as 7alpha-, 7beta-, 11-, 15alpha- and 16alpha- were successfully hydroxylated within this thesis by the use of different bacterial P450 enzymes and the enzymes’ selectivities varied depending of the applied substrates.