Reconstructing the hydrological change based on dendrohydrological data has important implications for understanding the dynamic distribution and evolution pattern of a given river. The widespread, long-living coniferous forests on the Altay Mountains provide a good example for carrying out the dendrohydrological studies. In this study, a regional composite tree-ring width chronology developed by Lariat sibirica Ledeb. and Picea obovata Ledeb. was used to reconstruct a 301-year annual （from preceding July to succeeding June） streamflow for the Haba River, which originates in the southern Altay Mountains, Xinjiang, China. Results indicated that the reconstructed streamflow series and the observations were fitting well, and explained 47.5% of the variation in the observed streamflow of 1957-2008. Moreover, floods and droughts in 1949-2000 were precisely captured by the streamflow reconstruction. Based on the frequencies of the wettest/driest years and decades, we identified the 19th century as the century with the largest occurrence of hydrological fluctuations for the last 300 years. After applying a 21-year moving average, we found five wet （1724-1758, 1780-1810, 1822-1853, 1931-1967, and 1986-2004） and four dry （1759-1779, 1811-1821, 1854-1930, and 1968-1985） periods in the streamflow reconstruction. Furthermore, four periods （1770-1796, 1816-1836, 1884-1949, and 1973-1997） identified by the streamflow series had an obvious increasing trend. The increasing trend of streamflow since the 1970s was the biggest in the last 300 years and coincided with the recent warming-wetting trend in northwestern China. A significant correlation between streamflow and precipitation in the Altay Mountains indicated that the streamflow reconstruction contained not only local, but also broad-scale, hydro-climatic signals. The 24-year, 12-year, and 2.2-4.5-year cycles of the reconstruction revealed that the streamflow variability of the Haba River may be influenced by solar activity and the atmosphere-ocean system.
