An approach to dealing with the contamination problem in the context of a simple model of a multi-process data base environment is described. We present an algorithm which, given a specification of a set of data transfers which are judged to have been possible contaminators, tracks the possible spread of contamination among processes and data blocks. This algorithm can be used as a diagnostic tool for recovery. Also, for environments in which it is feasible to rerun processes, we describe an algorithm which determines a recovery strategy. The strategy consists of processes to be rerun, blocks to be restored to a prior image, and appropriate prior images for these blocks.