Recent studies have shown that Plasmodium falciparum is sensitive to a purine salvage block at purine nucleoside phosphorylase (PNP) and that human PNP is a target for T-cell proliferative diseases. Specific tight-binding inhibitors might be designed on the basis of specific PNP transition state structures. Kinetic isotope effects (KIEs) were measured for arsenolysis of inosine catalyzed by P. falciparum and human purine nucleoside phosphorylases. Intrinsic KIEs from [1'-(3)H]-, [2'-(3)H]-, [1'-(14)C]-, [9-(15)N]-, and [5'-(3)H]inosines were 1.184 +/- 0.004, 1.031 +/- 0.004, 1.002 +/- 0.006, 1.029 +/- 0.006, and 1.062 +/- 0.002 for the human enzyme and 1.116 +/- 0.007, 1.036 +/- 0.003, 0.996 +/- 0.006, 1.019 +/- 0.005, and 1.064 +/- 0.003 for P. falciparum PNPs, respectively. Analysis of KIEs indicated a highly dissociative D(N)A(N) (S(N)1) stepwise mechanism with very little leaving group involvement. The near-unity 1'-(14)C KIEs for both human and P. falciparum PNP agree with the theoretical value for a 1'-(14)C equilibrium isotope effect for oxacarbenium ion formation when computed at the B1LYP/6-31G(d) level of theory. The 9-(15)N KIE for human PNP is also in agreement with theory for equilibrium formation of hypoxanthine and oxacarbenium ion at this level of theory. The 9-(15)N KIE for P. falciparum PNP shows a constrained vibrational environment around N9 at the transition state. A relatively small beta-secondary 2'-(3)H KIE for both enzymes indicates a 3'-endo conformation for ribose and relatively weak hyperconjugation at the transition state. The large 5'-(3)H KIE reveals substantial distortion at the 5'-hydroxymethyl group which causes loosening of the C5'-H5' bonds during the reaction coordinate.