Base on BIOME-BGC model and tree-ring data, a modeling study was conducted to estimate the dynamic changes of the net primary productivity ( NPP) of Abies faxoniana forest ecosystem in the Miyaluo Natural Reserve of western Sichuan of China (31. 04°~31. 92°N, 102. 58°~103. 07°E) . The sample site (31.63°N, 102. 96°E) of A. faxoniana in this study is located in the Jiabigou Ravine. In total, we collected 88 cores from 88 living trees of S. squamata in a treeline site (4000m above sea level) at the northern slope side. A total of 76 increment cores were eventually crossdated and selected for chronology development. The resulted forest NPP reconstruction, spanning 1788 ~ 2010 ( 233 years), could explain 44.8% the variance ( adjusted for degrees of freedom was 43. 8% ) contained in NPP data modeled by BIOME-BGC from the calibration period 1954 ~ 2008. The reconstruction indicated the annual NPP of Abies faxoniana forest in 1788~2010 fluctuated from 498. 66gC/m~2/a to 563. 65gC/m~2/a, with the mean value and standard deviation of 527. 2gC/m~2/a and 12. 45gC/m~2/a, respectively. The major decreasing episodes for forest NPP occurred during 1788 ~1811,1832 ~1844,1890 ~1928 and 1969-1993,whereas the distinct increasing episodes for forest NPP occurred during 1812 ~ 1831, 1845 ~1860,1929 ~ 1968 and 1994 ~2010. The reconstructed NPP in this study revealed similarities at inter-decadal frequencies with tree-ring reconstructed temperatures from nearby regions, as well as the delta~(18)O value of ice cores and recorded glacier fluctuations in the surrounding high mountains,which suggested the spatial fidelity of forest growth pattern deduced from our tree-ring data. Correlation and response analysis revealed that A. faxoniana forest NPP in this site was mostly governed by temperature from summer to autumn season ( especially for months of July and September), and the temperature in prior autumn and winter played a secondary role to explain the climate related variance in forest NPP series. In contrast, no consistent response pattern was found between forest NPP and precipitation, only precipitation in summer to early summer suggests the existence of negative impact on forest growth. Tree-ring data act as reliable proxy to predict the ecosystem dynamic change over ecosystem and regional scales, and could be used to validate the process-based ecosystem models and reconstruct the historical forest NPP during the past centuries.