The depth and spatial distribution of throughfall under the forest canopy can affect the hydrological (i.e., infiltration, runoff and evaporation) and ecological processes (i.e., tree growth, litter decomposition) of forest ecosystems. Therefore, both the evaluation of forest hydrological impacts and forest ecosystems management require an accurate description of the spatial distribution of throughfall. In order to interpret the spatial heterogeneity of throughfall in the plantation of Larix principis-ruprechtii, a 30 m × 30 m sized stand plot was set up in the Xiangshuihe watershed of the Liupan Mountains. Within this plot, 40 throughfall collectors, with a size of 230.58 cm², were evenly installed. The throughfall depth of 34 rainfall events was measured individually during the growth season (May-Oct.) of 2015. The gross rainfall in open field in the study period was 567.0 mm, and the throughfall depth in the plantation plot was 483.5 mm, accounting for 85.3% of gross rainfall. As rainfall increased, the throughfall ratio first increased and then tended to stabilize. A considerable spatial variation of throughfall was observed, and the coefficient of variance of throughfall first decreased and then became constant with increase in rainfall. The spatial distribution of throughfall was influenced by the tree characteristics canopy leaf area index and canopy thickness above throughfall collector, and the distance from throughfall collector to the nearest stem, and this influence varied with the depth of rainfall. When the rainfall was less than 10 mm, throughfall showed a significantly negative correlation with the leaf area index; when the rainfall varied within the range of 10-20 mm, throughfall showed a significant positive correlation with the distance to the nearest stem and a significantly negative correlation with canopy thickness. However, when the rainfall was more than 20 mm, throughfall showed no correlation with any of the tree characteristics. Considering that the depth of individual rainfall events is concentrated within 0-10 mm, at least 13 or 9 throughfall collectors with the size used here are required for a plot of 30 m×30 m if the error of measured throughfall should be controlled within 10% under confidence levels of 95% or 90%. At least 26 or 23 throughfall collectors are required when the error should be controlled within 5% under confidence levels of 95% or 90%.