Abstract: To investigate the response of wood density to climate change, wood density of Picea crassifolia trees at lower tree line in the middle Qilian Mountains, northwestern China was measured using Silviscan-3. Chronologies of annual, earlywood, latewood, maximum and minimum density were established by dendrochronological methods. Relationships of chronologies to monthly mean, maximum, minimum temperatures and monthly precipitation were analyzed. The results indicated that wood density of P.crassifolia trees positively correlated with temperatures and negatively correlated with precipitation. Annual and earlywood density significantly correlated with monthly mean temperatures in June to September, with monthly maximum temperatures in June, July and September, and with monthly minimum temperatures in October in previous year and in July and September in current year. Minimum density significantly correlated with temperature in June to August, with monthly maximum temperatures in June and July, and with monthly minimum temperatures in July. Annul density significantly correlated with precipitation in March. Earlywood and minimum density significantly correlated with precipitation in June. Latewood and maximum density did not show significant relationships with temperature and precipitation, latewood density was less sensitive to climate than earlywood density. Those results indicate that temperature has directly influence and precipitation has indirectly impact on wood density, temperature in summer is main climate factor influencing wood density of P.crassifolia in Qilian Mountains, northwestern China.