Existing studies pointed out that illiquidity of real estate operation in the long-term portfolio selection is caused mainly by three factors: non-i.i.d. characteristics of real estate return, high transaction cost and time-on-market (TOM). This paper explored a method of dynamic optimization of real estate investment in the long-term portfolio selection considering illiquidity risk especially from a view of high transaction cost. As a result it was found that the method of dynamic programming is useful for such a purpose as flexible simulation of illiquidcharacteristic dynamic system, and that level-holding of real estate is the best solution for real estate investment in some instances under the situation of risk-averting-investment. And also the method seems to have potential to incorporate non-i.i.d. characteristics and TOM.