In view of the medical service demand features faced by the present domestic medical institutions, i.e. mostly the walk-ins patient and the patient arrival surge, the idea of load balancing is introduced and a flexible outpatient scheduling method with multi-objective appointment optimization is proposed. According to the distribution of the actual walk-in load, the total load is balanced with flexible appointment load balance model in each period. Several appointment rules, queuing rules and reservation rates combinations are utilized to integrate a large number of decision schemes. The optimal solution under different scenarios is screened out by multi-attribute decision model of grey targets. The experiments show that the average wait time can be released to 77.2% comparing to other models. Application case show that resource use rate be improved by 66.7% than before, and the resource sensitivity is lower by using the proposed model.