This paper investigates the impact of allowing patient recovery in the operating room when no recovery bed is available. Three types of identical resources are considered: transporters, operating rooms and recovery beds. A fixed number of patients must be planned over a term horizon, usually one or two weeks. The surgery process is modelled as follows: each patient is transported from the ward to the operating theatre. Then the patient visits an operating room for surgery operation and is transferred to the recovery room. If no recovery bed is available, the patient wakes up in the operating room until a bed becomes available. The operating room needs to be cleaned after the patient's departure, before starting another operation. Finally, the patient is transported back to the ward after his recovery. We consider several criteria based on patients' completion times. We propose a Lagrangian relaxation-based method to solve this operating theatre scheduling problem. The efficiency of this method is then validated by numerical experiments. A comprehensive numerical experiment is then performed to quantify the benefit of allowing patient recovery in operating rooms. We show that the benefit is high when the workload of the recovery beds is high.