The process of rollback and recovery is a form of redundancy introduced into data base systems to enhance the integrity and consistency of the data. Rollback and recovery consists of periodical checkpointing (storing a copy of relevant data) and keeping a chronological record of transactions which modify the data base on an audit trail. When an error is detected in the data base, the data base copy made at the most recent checkpoint is loaded and all update transactions stored on the audit trail since the last checkpoint are reprocessed and applied to this copy. This thesis analyzes the process of rollback and recovery for multicopy files in a distributed data base. Models are developed which find the optimal time between checkpoints. The practical considerations concerning a control of rollback and recovery in a distributed data base environment are also presented.