Reduction of NDPs by murine ribonucleotide reductase (mRR) requires catalytic (mR1) and free radical-containing (mR2) subunits and is regulated by nucleoside triphosphate allosteric effectors. Here we present a new, comprehensive, and quantitative model for allosteric control of mRR enzymatic activity based on molecular mass, ligand binding, and enzyme activity studies. In this model, nucleotide binding to the specificity site (s-site) drives formation of an active R1(2)R2(2) dimer, ATP or dATP binding to the adenine-specific site (a-site) results in formation of an inactive tetramer, and ATP binding to the newly described hexamerization site (h-site) drives formation of active R1(6)R2(6) hexamer. In contrast, an earlier phenomenological model [Thelander, L., and Reichard, P. (1979) Annu. Rev. Biochem. 67, 71-98] (the "RT" model) ignores aggregation state changes and mistakenly rationalizes ATP activation versus dATP inhibition as reflecting different functional consequences of ATP versus dATP binding to the a-site. Our results suggest that the R1(6)R2(6) heterohexamer is the major active form of the enzyme in mammalian cells, and that the ATP concentration is the primary modulator of enzyme activity, coupling the rate of DNA biosynthesis with the energetic state of the cell. Using the crystal structure of the Escherichia coliR1 hexamer as a model for the mR1 hexamer, a scheme is presented that rationalizes the slow isomerization of the tetramer form and suggests an explanation for the low enzymatic activity of tetramers complexed with R2. The similar specific activities of R1(2)R2(2) and R1(6)R2(6) are inconsistent with a proposed model for R2(2) docking with R1(2) [Uhlin, U., and Eklund, H. (1994) Nature 370, 533-539], and an alternative is suggested.