The transient decay reaction kinetics of 1,1,2,2- ¹H₄- and 1,1,2,2-²H₄-aminoethanol-generated Co^II-substrate radical pair catalytic intermediate in ethanolamine ammonia-lyase (EAL) have been measured by using time-resolved, X-band continuous-wave electron paramagnetic resonance (EPR) spectroscopy in frozen aqueous solution from 190 to 223 K. The decay is biexponential at temperature T<214 K (¹H) or <210 K (²H), with fast and slow phase first-order rate constants k_obs,f and k_obs,s, respectively. The decay becomes monoexponential at temperature T≥214 K, with rate constant k _obs,m. The kobs,f and kobs,m values adhere to the same linear relation on a lnk versus T^-1 (Arrhenius) plot, and therefore represent the same mechanism, which is proposed to be the native forward reaction of the substrate radical through the radical rearrangement step. The ¹H/²H isotope effect (IE) on k_obs,f of 1.4±0.1 at 190≤T≤207 K is assigned to an α-secondary hydrogen kinetic IE on the rearrangement step. The k_obs,s values obey a different Arrhenius relation, and display an inverse kinetic IE (0.8±0.1). The slow decay phase is proposed to be associated with the forward reaction, but with a different rate determining step. The ¹H/²H IE on k_obs,m increases continuously at T>210 K, to 2.1±0.1 at 223 K. A three-state (substrate radical, product radical, diamagnetic products), two-step [rearrangement, and subsequent hydrogen atom transfer, (HT)] model is used to generate a consistent fit to the temperature dependence of the _kobs,f, k_obs,m values and IEs at low temperature with k_cat values and IEs at 277 K (IE=5.5) and 293 K (IE=7.8). The model shows that the four decade-old paradox of ¹H/²H and ¹H/³H IEs in EAL, and the temperature dependent IE, are caused by a significant negative activation entropy for the HT step, relative to rearrangement. The bifurcation of the decay kinetics at 207 HT step, relative to rearrangement. The bifurcation of the decay kinetics at 207< T<214 K is addressed by measuring the detailed (1 K intervals) temperature dependence of samples prepared with only slow phase population. The steep lnk versus T^-1 dependence is discontinuous with the fast and slow phase relations. The origin of the kinetic bifurcation is proposed to arise from a protein dynamical transition, which is coupled to the core adiabatic reaction in EAL.