It is shown that a simple kinetic model can account for existing data both on the formation of the native defect EL2 in the temperature range 644–800 °C in GaAs samples from which EL2 was eliminated by a 1200 °C anneal/quench and on the disappearance of EL2 during anneals in the temperature range 1000–1200 °C. Our analysis suggests that EL2 consists ofVGabound to an unidentified ‘‘kernel’’ which, if not actually stable at temperatures up to 1200 °C, forms relatively rapidly at the lower temperatures and dictates the final concentration of EL2 in the sample. The change in enthalpy involved in the capture or release ofVGaby the kernel is estimated to be 5.6 eV.