Within past ten years, the importance of cellular automata (CA) models of both recrystallization and grain boundary migration has been recognized. It can be shown that CA provides a computationally efficient mathematical framework for simulations of physically relevant models of the above processes. Generally, CA works in such models with a vast number of identical copies of several generic processes - as recrystallization, nucleation, grain boundary migration, etc. - which interact locally and results in a complex global response. The attention is focused to some important aspects of recrystallization, grain growth and CA modelling. Finally, possible future development of more physically relevant models are briefly outlined.