Natural and laboratory ESR dose response curves (DRCs) of [AlO4/h]0 and [TiO4/M+]0 were investigated for samples of quartz from the Luochuan loess-palaeosol master section, Chinese Loess Plateau. The natural and laboratory DRCs show a clear divergence above ~1000 Gy, with much lower D0 values and saturation levels observed for the natural DRCs, which is in agreement with the previous study by Tsukamoto et al. (2018). Young (<15 ka) samples from Luochuan and Jingbian –another site of the Chinese Loess Plateau, together with two modern samples of Chinese loess, were used to investigate the residual signals of [AlO4/h]0 and [TiO4/M+]0 centres. Our results are in line with published studies and show that the significant residual signals corresponding to several tens to hundreds of Gy are present in both Al and Ti centres. These need to be taken into account before laboratory DRC construction. ESR pulse annealing experiments performed on samples irradiated with different doses show an apparent dose-dependent thermal instability of [AlO4/h]0 and [TiO4/M+]0, with the signals for higher doses decaying faster with increasing temperature. We attribute the change in D0 with preheat reported in Tsukamoto et al., 2018, as well as the difference between laboratory and natural DRCs, to this apparent dose-dependent thermal instability of the signals. The saturation level of the natural DRC, being the result of reaching the equilibrium between filling of the traps and emptying them due to thermal decay, is therefore additionally affected at higher doses, due to the increased thermal instability. The inability to recreate in the laboratory the same response to irradiation as the one observed in nature questions the accuracy of dating samples beyond 1000 Gy.