The class I ribonucleotide reductases (RNRs) are composed of two homodimeric subunits: R1 and R2. R2 houses a diferric-tyrosyl radical (Y•) cofactor. Saccharomyces cerevisiae has two R2s: Y2 (β 2 ) and Y4 (β′ 2 ). Y4 is an unusual R2 because three residues required for iron binding have been mutated. While the heterodimer (ββ′) is thought to be the active form, several rnr4Δ strains are viable. To resolve this paradox, N-terminally epitope-tagged β and β′ were expressed in E. coli or integrated into the yeast genome. In vitro exchange studies reveal that when apo-(His 6 )-β 2 ( His β 2 ) is mixed with β′ 2 , apo-His ββ′ forms quantitatively within 2 min. In contrast, holo-ββ′ fails to exchange with apo-His β 2 to form holo-His ββ and β′ 2 . Isolation of genomically encoded tagged β or β′ from yeast extracts gave a 1:1 complex of β and β′, suggesting that ββ′ is the active form. The catalytic activity, protein concentrations, and Y• content of the rnr4Δ and wild type (wt) strains were compared to clarify the role of β′ in vivo. The Y• content of rnr4Δ is 15-fold less than that of wt, consistent with the observed low activity of rnr4Δ extracts (<0.01 nmol min −1 mg −1 ) versus wt (0.06 ± 0.01 nmol min −1 mg −1 ). FLAG β 2 isolated from the rnr4Δ strain has a specific activity of 2 nmol min −1 mg −1 , similar to that of reconstituted apo-His β 2 (10 nmol min −1 mg −1 ), but significantly less than holo-His ββ′ (~2000 nmol min −1 mg −1 ). These studies together demonstrate that β′ plays a crucial role in cluster assembly in vitro and in vivo and that the active form of the yeast R2 is ββ′.Ribonucleotide reductases (RNRs) 1 catalyze the conversion of ribonucleotides to deoxyribonucleotides, providing the monomeric precursors for DNA replication and repair (1). The class I RNRs are composed of a large subunit, R1, and a small subunit, R2. R1 contains the site of nucleotide reduction and the allosteric effector binding sites that control the rate and the specificity of nucleotide reduction. R2 houses the diferric-tyrosyl radical † D.L.P. and A.D.O. were supported in part by the NIH training grant 5T32 CA 09112-28. J.S. acknowledges support of the NIH (GM29595). M.H. acknowledges support of the NIH (CA095207) and the ACS (0305001GMC).