Hydrogen peroxide is involved in a variety of enzyme catalysis as an oxidant or toxic by-product. Thereby, attenuation of the H2O2-driven oxidative stress is one of the key issues for preparative biocatalysis. Here, a rational approach to improve the robustness of enzymes, in particular, Baeyer-Villiger monooxygenases (BVMOs) against H2O2 was investigated. The enzyme access tunnels, which may serve as exit paths for H2O2 from the active site to the bulk, were predicted by using the CAVER and/or protein energy landscape exploration (PELE) software for the phenylacetone monooxygenase variant (PAMO_C65D) from Thermobifida fusca and the BVMO from Pseudomonas putida KT2440. The amino acid residues, which are susceptible to oxidation by H2O2 (e. g., methionine and tyrosine) and located in vicinity of the predicted H2O2 migration paths, were substituted with less reactive or inert amino acids (e. g., leucine and isoleucine). This led to design of the H2O2-resistant enzyme variants, which became robust biocatalysts for synthetic applications. For instance, the H2O2-resistant P. putida BVMO reached turnover numbers of 4,100 for the BV oxygenation of 4-decanone, which is 2.8-fold greater than the parent enzyme. Moreover, the H2O2-resistant P. putida BVMO allowed 2-fold enhancement in titer of 9-(nonanoyloxy)nonanoic acid (8) formation in a cascade fatty acid biotransformation. Therefore, it was assumed that the CAVER/PELE-based H2O2 migration path engineering represents an efficient rational design approach to improve not only oxidative stability but also biotransformation performance of the H2O2-forming or utilizing enzymes (e. g., BVMOs, oxidases, and peroxidases). (Figure presented.).