Cytochromes P450 have been recently identified as a promising class of biocatalysts for mediating C-H aminations via nitrene transfer, a valuable transformation for forging new C-N bonds. The catalytic efficiency of P450s in these non-native transformations is however significantly inferior to that exhibited by these enzymes in their native monooxygenase function. Using a mechanism-guided strategy, we report here the rational design of a series of P450(BM3)-based variants with dramatically enhanced C-H amination activity acquired through disruption of the native proton relay network and other highly conserved structural elements within this class of enzymes. This approach further guided the identification of XplA and BezE, two "atypical" natural P450s implicated in the degradation of a man-made explosive and in benzastatins biosynthesis, respectively, as very efficient C-H aminases. Both XplA and BezE could be engineered to further improve their C-H amination reactivity, which demonstrates their evolvability for abiological reactions. These engineered and natural P450 catalysts can promote the intramolecular C-H amination of arylsulfonyl azides with over 10 000-14 000 catalytic turnovers, ranking among the most efficient nitrene transfer biocatalysts reported to date. Mechanistic and structure-reactivity studies provide insights into the origin of the C-H amination reactivity enhancement and highlight the divergent structural requirements inherent to supporting C-H amination versus C-H monooxygenation reactivity within this class of enzymes. Overall, this work provides new promising scaffolds for the development of nitrene transferases and demonstrates the value of mechanism-driven rational design as a strategy for improving the catalytic efficiency of metalloenzymes in the context of abiological transformations.