To investigate the differences in the climatic signals in stable-carbon isotopic composition captured by averaging series from individual trees versus raw wood of trees pooled prior to analysis, we analyzed two groups of Chinese pine (Pinus tabulaeformis Carr.) from the Nan Wutai region of the Qinling Mountains, China. One group included three trees that were analyzed separately, and the other group comprised four other trees that were pooled prior to preparation and analysis. All δ13C series were positively correlated (r = 0.50–0.58, p < 0.0001) for the period AD 1901–2003. After removing the effects of changing δ13C of atmospheric CO2, correlations between the meteorological data and all individual and pooled discrimination (∆13C) series revealed significant negative responses to temperature for several specific months and for mean January to September (TJ–S) temperature. We used a “numerical mix method” (NMM, equivalent to unweighted mean), by averaging individual ∆13C series (NPS1+NPS2+NPS3), to generate a new series that more strongly correlated to climate series TJ–S (r = − 0.67, p < 0.0001). This time interval from January through September (J–S) includes the growing season and months prior to the growing season, but the temperature prior to the growing season may provide energy necessary for timely initiation of growth. Thus, the mean TJ–S is significant for plant growth and is consistent with the tree physiology in this region. Our results suggest that the numerical mix method with tree-ring stable isotope data from three trees provides a series quite satisfactory for climatic reconstruction. The relationship of the numerical mix model ∆13C with temperature was stronger than for the pooled series, suggesting numerical mixing of series can be more effective than raw wood sample pooling at least according to the trees in this study.