Considering multiple transmit antennas in each distributed antenna unit (DAU), two power allocation (PA) schemes are proposed for energy efficiency (EE) maximization for downlink distributed multiple-input single-output (DMISO) systems with orthogonal frequency-division multiplexing (OFDM) over frequency-selective fading channels, where the power constraints for individual antenna units are addressed. The optimization problem for the maximization of the EE subject to per-antenna maximum power constraints is formulated. By means of linear programming, the optimization is simplified to as if for a DMISO system whose DAUs use a single antenna corresponding to the largest channel-gain-to-noise ratio (CGNR) for transmission. Using the block coordinate descent (BCD) method, an iterative optimal PA scheme to the simplified optimization problem is derived, where an efficient procedure for determining the number of effective subcarriers and the optimized PA is developed. Since the optimal scheme needs iterative calculations, a closed-form suboptimal PA scheme is also derived by sorting the total CGNR and using the principle of the BCD method. Interestingly, this suboptimal scheme has small performance loss in comparing with the optimal scheme, and its relative EE loss is decreased with the number of subcarriers. Moreover, these two schemes include the ones with single transmit antenna for distributed antenna systems as special cases. Computer simulations verify the effectiveness of the two proposed schemes, and the proposed optimal one can obtain the same performance as the existing optimal scheme for DMISO-OFDM but with lower complexity.