In heavily doped silicon, dopant aggregation leads to a reduction in the electrically active concentration. For thermal cycles typical of very large scale integrated circuit fabrication, the kinetics of this precipitation process can play a critical role in determining the active doping levels and thus device electrical characteristics. In this work, we develop a general kinetic precipitation model which explicitly considers the time evolution of the precipitate size density, and is thus able to account for a broad range of behavior. We compare our model to previously published experimental data for arsenic and phosphorus activation/deactivation in silicon over a wide temperature range (300 to 800 degrees C), and find that it can account accurately for the experimental observations, including reverse annealing following a temperature step.