In this paper we present an emergency repair model. We model a situation where failed parts arrive at a repair shop according to a Poisson process. If the stock on hand of serviceable spare parts exceeds a given emergency trigger level, the failed part is sent into normal repair. Otherwise the failed part is sent into an alternative emergency repair channel. Upon arrival failed parts are exchanged for serviceable spare parts, if available. Otherwise the demand is backordered. A backorder is filled with the first part that becomes available from either one of the repair processes. The repair times (including transportation time) are assumed to be exponentially distributed. We analyze the impact of using emergency repair on two service measures: the fraction of demand that is satisfied from stock on hand and the expected duration of a backorder. We calculate the initial stock level and the emergency trigger level that minimize the total cost for a given emergency repair rate. The proposed policy is compared with two other policies: (1) emergency repair is not used and (2) emergency trigger level is zero. We find that significant service improvements and cost reductions can be obtained by using our emergency repair policy. We also present simulation results showing that the distribution of the repair times has a negligible effect on the relevant service measures. Finally, we compare our results with the results of an emergency supply model by Muckstadt and Thomas [1] Again we observe significant cost reductions when using our policy.