Global climate change is characterized by variations in precipitation patterns and frequency of drought events. Extreme drought events may lead to large-area tree death and thus changes in the geographical distribution of vegetation,of which the underlying mechanisms are still unclear. It has been postulated that carbon starvation may be responsible for the drought-induced tree death, which, however, needs to be tested with a well-designed controlled experiment. In this study, we conducted a rainfall exclusion experiment on Dahurian larch (Larix gmelinii Rupr.) plantation during the growing season of 2012 and examined the concentrations of non-structural carbohydrates (NSC) in twigs and needles of the larch. The experiment included three treatments: 100% rainfall exclusion (100% RE), 50% rainfall exclusion (50% RE) and control (CK, i.e., ambient rainfall). Our objectives were to (1) compare the concentration of NSC components in the needles and twigs among the treatments; and (2) explore relationships between NSC concentrations in needles or twigs and soil moisture. Three trees were randomly selected for each treatment. The twigs and needles from upper, mid and lower crowns in the southern direction were monthly sampled and mixed,respectively,for each tree from early June to late August of 2012. A modified phenol-sulphuric method was used to determine the concentrations of soluble sugars and starch. For a specific sample, the total concentration of NSC was obtained by by summing the concentrations of the sugars and starch. There was no significant effect of the treatments on the total NSC concentrations in both needles and twigs (P > 0.005). The mean concentrations of total NSC in needles varied from 9.45 mg/g to 14.12 mg/g, and those in twigs ranged between 7.72 mg/g and 9.26 mg/g. However, the concentration of soluble sugars in needles differed significantly among the treatments (P < 0.05). The sugar concentration in needles was highest for the 100% RE treatment (8.980.31) mg/g, followed by the 50% RE treatment (8.450.13) mg/g, and the lowest for the CK (7.730.32) mg/g. In contrary, the starch concentration in needles ranked as the CK (2.990.22)mg/g > 50% RE (2.680.32)mg/g > 100% RE (2.63 0. 17) mg/g. The ratio of soluble sugars to starch in the needles ranked as the CK (2.27) < 50% RE (2.51) < 100% RE (3.70). There was a significant negative relationship between soil moisture and concentration of soluble sugars or ratio of sugar to starch in needles (P < 0.05). The concentration of starch in needles tended to increase with soil moisture increasing, but their correlation was insignificant (P > 0.05). There were no significant correlations between soil moisture and the total NSC concentration in needles, and between soil moisture and the concentrations of total NSC or component concentrations in twigs (P > 0.05). This study illustrated that the short-term drought stress had no significant effect on the total NSC concentration,but resulted in a transfer of starch into soluble sugar that maintained tree metabolic activities and survival during the drought event.