Distributed multipoint systems (DMS) are important and timely for the move to future broadband wireless communication systems. Traditional studies on DMS have mainly focused on the issues with the spatial division multiple access such as precoding techniques, which only consider a narrowband case. This paper addresses the downlink radio resource management of the orthogonal frequency division multiple access DMS (OFDMA-DMS), including power allocation between users or subcarriers,and distributed antenna selection. Signal models with incoherent and coherent transmitters are built. To maximize the system throughput, for the incoherent transmitter case, a strategy based on the iterative water-filling power allocation is proposed to approach the optimality. As for the coherent case, where coherent additions of the signal could occur at the users, the problem is transformed into an integer programming which is solvable. Numerical results show that the gain from the coherent transmitter is promising. And to achieve a near-optimal solution, only part of the DA ports will be used, which have better channel conditions.