Dynamic channel allocation (DCA) has been seen as a promising method for improving quality and capacity in cellular networks. In theoretical studies very significant quality and capacity improvements has been achieved. However, in real systems DCA performance is limited by several factors so theoretical limits might not be reached. To exploit full possibilities of TDD system efficient DCA algorithms are needed to guarantee quality of service in rapidly changing channel conditions and in mixed traffic scenarios. In this paper UTRA TDD down-link system capacity is analyzed with different DCA schemes. Dynamic system simulations are carried out to find out how the different elements in the DCA process impact the downlink system capacity. Effect of random and downlink transmission power based channel allocation with and without slow DCA information is studied in presence of signaling delays.