A new atomistic approach to Si device process simulation is presented. It is based on a Monte Carlo diffusion code coupled to a binary collision program. Besides diffusion, the simulation includes recombination of vacancies and interstitials, clustering and re‐emission from the clusters, and trapping of interstitials. We discuss the simulation of a typical room‐temperature implant at 40 keV, 5×1013 cm−2 Si into (001)Si, followed by a high temperature (815 °C) anneal. The damage evolves into an excess of interstitials in the form of extended defects and with a total number close to the implanted dose. This result explains the success of the ‘‘+1’’ model, used to simulate transient diffusion of dopants after ion implantation. It is also in agreement with recent transmission electron microscopy observations of the number of interstitials stored in (311) defects.