The grid-based Eulerian/particle-based Lagrangian method of analysis was employed to verify the formation of NO, NO2, and N2O and their interaction in a rapeseed methyl ester (RME) turbulent nonpremixed flame. In the Eulerian part, the hybrid algorithm Reynolds-averaged Navier-Stokes (RANS)/probability density function (PDF) was utilized to resolve the vector and scalar fields across the flame. In the Lagrangian part, stochastic tracking of trajectories of liquid droplets was performed to model the dispersion of particles due to the turbulence. The instantaneous concentrations of the NOX species was related to turbulence based on the self-similarity of ten diffusion flames (i.e., flamelet). Discrete ordinates (DO) was exploited for modeling of heat radiation and increase of the accuracy of the computational simulation in prediction of thermal NO. The numerical analysis has a great ability in prediction of NOX in boundaries of the flame; however, it shows some weakness in giving the concentration of NOX species in postflame. Results reveals that NOX species (NO, NO2 and N2O) start to evolve in flame boundaries where the time scales of flow are large and combustion is almost complete. The self-similarity observed in the evolutions of the NOX decays in the postflame. In the absence of prompt NO formation, the rate of the conversion of NO2 and N2O to NO is significantly higher than thermal NO formation although the thermal NO is main contributor to the turbulent flame. Among the elemental reaction during the thermal NO formation, the rate of underestimation elemental reaction (N + OH double left right arrow NO + H) is higher than other two elemental reactions in RME turbulent flame.