Flow-of-funds(FOF)analysis is an important part of SNA (System of National Accounts)．However，the collecting of a large amount of data causes the problem of long time lag for flow-of-funds table of physical transactions in many countries．The common methods used to update FOF tables were based on the assumption that transaction structures of the base and forecast periods remain constant．Yet this assumption does not always hold，especially in countries which experience significant changes in the economic structure． To solve this problem，a more flexible framework to forecast FOF table of physical transaction is proposed． Firstly，the elements in the physical transaction FOF table are divided into four categories． Secondly，based on the constraints that must be met by the elements in the table，the FOF table makes predictions in the target year by establishing a mathematical model to relieve the relevant constraints and by using the dynamic trend of historical data．The validity and stability of the proposed method are verified by a simulation experiment．Finally，the proposed method is applied to China's 1992－2014 FOF table and satisfactory results are achieved.