The wide-gap semiconductor ZnO has a band gap of ’3:3 eV at room temperature and is of current interest due to its potential optoelectronic applications. Since undoped ZnO is known to be grown under fairly high residual n-type, the trial of the p-type doping generally tends to result in an observation of the photoluminescence (PL) of donor–acceptor pair (DAP) bands. The optical properties of the DAP emissions have not been fully understood. In a previous report, we described the growth and the steady-state PL properties of ZnO:N epitaxial layers grown by repeated temperature modulation (RTM) epitaxy. Evidence of p-type conductivity for RTM-grown ZnO:N has been accumulated from a variety of experimental results, such as the Hall-effect and electroluminescence measurements in a p–n juction structure. For the analysis of PL peak energy and excitation-induced peak shift on p-type ZnO, we took the effect of donor wavefunction overlapping and the fact of impurity distribution inhomogeneity into account by adopting the fluctuation theory developed by Kuskovsky et al. Using this theory, we were able to consistently explain the excitation-intensity dependences of the PL properties of ZnO:N grown with three different techniques and conditions. In this short communication, we report on the analysis of the time-resolved PL (TRPL) spectra taken for the RTMgrown sample at an intermediate doping level to experimentally detemine the transition probability of the DAP (being W0) by using all the other material parameters that have been deduced in our previous work. It should be noted that the recombination dynamics of the DAPs has been studied for ZnO:N within the low-doping regime, where the approach of Thomas, Hopfield, and Augustyniak (THA) is still valid. The doped epilayer was grown by the RTM epitaxy on a lattice-matched ScAlMgO4 (SCAM) substrate. 7) The detailed methods of fabrication have been given elsewhere. Secondary-ion mass spectroscopy (SIMS) yielded a total nitrogen concentration [N] of 6 10 cm . Our ZnO:N sample dominantly showed the DAP band (D being a shallow donor) with zero-phonon energies of 3:235 eV, a donor activation energy of ED 1⁄4 60meV, and an acceptor activation energy of EA 1⁄4 171meV at 5K. These have been already determined from the energy of the bound exciton line (unchanged from that for undoped samples) and the excitation-intensity dependence of the PL. In addition, the energy of EA has turned out to be almost independent of the nitrogen concentration. For the time-resolved experiments, pulsed excitation was provided by the frequency-doubled beam of a picosecond mode-locked Ti:sapphire laser (360 nm, 2 ps). Here, we focus our attention to experimental total-lightdecay data (a spectrally integrated PL decay curve), because one can fit to the data with a smaller number of running parameters than the case of spectrally resolved decay curves, i.e., EA, ED, and ðNA NDÞ (acceptor and donor concentrations). Irrespective of the model adopted, in both cases of the theories of THA and of the fluctuation, the intensity IðtÞ of light emitted at time t is