A concept of the anthropogenic origin of the current global climate warming assumes that growth of concentration of the atmospheric carbon dioxide and other greenhouse gases is of great concern in this process. However, all earlier performed analyses of the Antarctic ice cores, covering the time interval of several glacial cycles for about 1 000 000 years, have demonstrated that the carbon dioxide concentration changes had a certain lag relative to the air temperature changes by several hundred years during every beginning of the glacial terminations as well as at endings of interglacials. In contrast to these findings, a recently published careful analysis of Antarctic ice cores (Parrenin et al., 2013) had shown that both, the carbon dioxide concentration and global temperature, varied almost synchronously during the transition from the last glacial maximum to the Holocene. To resolve this dilemma, a special technique for analysis of the paleoclimatic time series, based on the wavelets, had been developed and applied to the same carbon dioxide concentration and temperature time series which were used in the above paper of Parrenin et al., 2013. Specifically, a stack of the Antarctic δ18O time series (designated as ATS) and the deuterium Dome C – EPICA ones (dD) were compared to one another in order to: firstly, to quantitatively estimate differences between time scales of these series; and, secondly, to clear up the lead–lag relationships between different scales variations within these time series. It was found that accuracy of the mutual ATS and dD time series dating lay within the range of 80–160 years. Perhaps, the mutual dating of the temperature and carbon dioxide concentration series was even worse due to the assumed displacement of air bubbles within the ice. It made us to limit our analysis by the time scales of approximately from 800 to 6000 years. But it should be taken into account that any air bubble movement changes the time scale of the carbon dioxide series as a whole. Therefore, if a difference between variations in any temperature and the carbon dioxide time series is found to be longer than 80–160 years, and if these variations are timescale‑dependent, it means that the bubble displacements are not essential, and so these advancing and delays are characteristic of the time series being compared. Our wavelet‑based comparative and different‑scale analysis confirms that the relationships between the carbon dioxide concentration and temperature variations were essentially timescale‑dependent during the transition from the last glacial maximum to the Holocene. The carbon dioxide concentration variations were ahead of the temperature ones during transition from the glacial maximum to the Boelling – Alleroud warming as well as from the Young Drias cooling to the Holocene optimum. However, the temperature variations were ahead during the transition from the Boelling – Alleroud warming to the Young Drias cooling and during the transition from the Holocene optimum to the present‑day climate.