In a conventional class I ribonucleotide reductase (RNR), a diiron(II/II) cofactor in the R2 subunit reacts with oxygen to produce a diiron(III/IV) intermediate, which generates a stable tyrosyl radical (Y*). The Y* reversibly oxidizes a cysteine residue in the R1 subunit to a cysteinyl radical (C*), which abstracts the 3'-hydrogen of the substrate to initiate its reduction. The RNR from Chlamydia trachomatis lacks the Y*, and it had been proposed that the diiron(III/IV) complex in R2 directly generates the C* in R1. By enzyme activity measurements and spectroscopic methods, we show that this RNR actually uses a previously unknown stable manganese(IV)/iron(III) cofactor for radical initiation.