Efficient batching for repair is essential to minimizing the 'equipment downtime' experienced by end-users when a repairable product fails. In this paper, the batching problem for a repair shop with limited spares and finite capacity is formulated and batching policies are developed. The performance of these policies is examined in a simulation model of a hypothetical repair shop. A policy called 'SP-WBPT' performed very well in all the environments tested. In this policy, the size of the batch is the number of failed parts of that part type waiting to be repaired. Batch selection is based onthe shortest batch processing time per part, weighted by the number of available spares.