By consideration of relative rates of diffusion and of propagation, it is shown that, in some bulk polymerizations, there exists a certain weight fraction of polymer (wp) beyond which initiator efficiency (f) must rapidly decrease. At this critical wp, the two free radicals formed from initiator decomposition are immobilized by propagation faster than they can diffuse apart; consequently, their likely fate is to become trapped in close proximity to each other and undergo geminate recombination. A quantitative theory for the onset of this phenomenon is derived. This effect explains why some bulk polymerizations are extremely slow at high conversions while the corresponding emulsion polymerizations are not; further, it removes apparent discrepancies between kp values directly measured from ESR in emulsion polymerizations and those estimated from bulk kinetics. Bulk kinetic data are employed to give the variation of f with wp for a range of initiator types and concentrations; these f(wp) are shown to be consistent with the proposed theory. This suggests that the common assumption that f is independent of conversion, while kp drops rapidly at the glass transition, is seriously in error. Instead, it appears that kp changes slowly beyond the onset of diffusion-controlled propagation while f drops dramatically beyond a conversion that depends on a number of chemical and physical properties of the system.