(6–4) Photolyases ((6–4) PLs) are ubiquitous photoenzymes that use the energy of sunlight to catalyze the repair of carcinogenic UV-induced DNA lesions, pyrimidine(6–4)pyrimidone photoproducts. To repair DNA, (6–4) PLs must first undergo so-called photoactivation, in which their excited flavin adenine dinucleotide (FAD) cofactor is reduced in one or two steps to catalytically active FADH− via a chain of three or four conserved tryptophan residues, transiently forming FAD•−/FADH− ⋯ TrpH•+ pairs separated by distances of 15 to 20 Å. Photolyases and related photoreceptors cryptochromes use a plethora of tricks to prevent charge recombination of photoinduced donor–acceptor pairs, such as chain branching and elongation, rapid deprotonation of TrpH•+ or protonation of FAD•−. Here, we address Arabidopsis thaliana (6–4) PL (At64) photoactivation by combining molecular biology, in vivo survival assays, static and time-resolved spectroscopy and computational methods. We conclude that At64 photoactivation is astonishingly efficient compared to related proteins—due to two factors: exceptionally low losses of photoinduced radical pairs through ultrafast recombination and prevention of solvent access to the terminal Trp3H•+, which significantly extends its lifetime. We propose that a highly conserved histidine residue adjacent to the 3rd Trp plays a key role in Trp3H•+ stabilization.