This study explored the effect of radical size, chain length, and mass on the radical cage effect. Radical cage pairs of the type [(RCp)(CO)(3)M-., M-.(CO)(3)(CpR)] (M = Mo or W; CpR = various substituted cyclopentadienyl ligands) were generated by photolysis (lambda = 540 nm) of the metal-metal bends in (RCp)(2)M-2(CO)(6) The cage recombination efficiencies (denoted as F-cP) for the radical cage pairs were obtained by extracting them from quantum yield measurements for the photoreactions with CCl4 (a metal-radical trap) as a function of solvent system viscosity. For the series of molecules (R3SiOCH2CH2Cp)(2)Mo-2(CO)(6), (R = Me, i-Pr, n-Pr, n-Hx), the F-cP values were linearly proportional to mass(1/2)/radius(2), in agreement with the predictions of Noyes＇ cage effect theory. It is also demonstrated that the difference in the cage recombination efficiencies between the [(MeCp)(CO)(3)Mo-., Mo-.(CO)(3)(CpMe)] and [(MeCp)(CO)(3)W-., W-.(CO)(3)(CpMe)] cage pairs cannot be ascribed to the different masses of the radicals. Rather, the difference is shown to be attributable to differences in the metal-metal bond energies or to differences in the spin-orbit coupling. In another comparison, F-cP at any viscosity for [(MeCp)(CO)(3)Mo-., Mo-.(CO)(3)(CpMe)] was shown to be identical to that of [Cp*(CO)(3)Mo-., Mo-.(CO)(3)Cp*] (Cp* = eta(5)-C5Me5) in tetrahydrofuran (THF)/tetraglyme solution. Rotation of the MeCp ring is fast compared to the time scale of diffusive separation (k(dP)) and radical recombination (k(cP)), and hence the effective volumes of the radicals in the solvent cage are nearly identical, which leads to similar F-cP values.