This study uses a decomposed queuing network model to determine optimal beds to serve an emergency department where patients have low tolerance for delay. M/M/∞ and M/M/C of queuing models were used to analyse the data obtained from LUTH Emergency Department (ED) for a period of 18 months. The findings reveal that the ED requires at least 64 beds to provide efficient emergency services to the existing population. It was suggested that improving patient flow and reducing waiting time for patient can be achieved through the following ways: (1) managing the arrival rate by different appointment methods such as online appointment management system to spread arrivals and call in appointment system; (2) increasing the number of servers; (3) optimizing the service rate.