Throughfall is the main form of rainfall redistribution, and it is crucial for soil moisture supply and vegetation growth in arid and semi-arid regions. Most previous studies focused on the throughfall of forests. The mechanisms that influence shrub throughfall, especially the effects of canopy characteristics on throughfall, need further quantitative study. Furthermore, there is little research on the spatial heterogeneity and aggregation effects of shrub throughfall (i.e., throughfall amount was larger than precipitation amount). In this study, the distribution of throughfall in eight directions and vegetation factors of a typical xerophytic shrub (Caragana korshinskii), as well as the rainfall characteristics in the Liudaogao catchment in the Loess Plateau of China, were measured over the year 2016. The variations in throughfall amount and percentage, spatial variability, and the aggregation effects of throughfall were investigated in detail. The major rainfall characteristics and vegetation factors affecting throughfall were identified, and the corresponding quantitative relationships were established. The total precipitation and throughfall amount during the study period were 509.68 mm and 439.51 mm, respectively, and the throughfall accounted for 86.23% of rainfall. The average values of throughfall amount, throughfall percentage, and coefficient of spatial variability of throughfall under rainfall events were 11.88 mm, 75.71%, and 21.80%, respectively. The throughfall amount was significantly correlated with precipitation amount, rainfall duration, I₁₀ (maximum rainfall intensity in 10 min), and I₃₀ (maximum rainfall intensity in 30 min) (P < 0.05), and it was mainly determined by precipitation amount. The throughfall amount increased linearly with precipitation amount (R² =0.99), and the threshold of precipitation to produce throughfall was 1.06 mm. The throughfall percentage and spatial variability of throughfall were significantly correlated with precipitation amount, rainfall duration, I₅ (maximum rainfall intensity in 5 min), I₁₀, and I₃₀ (P < 0.05), and they were mainly affected by precipitation amount and rainfall intensity. There was no significant correlation between throughfall and average rainfall intensity. Throughfall percentage increased logarithmically with precipitation amount and I₃₀ (R² =0.71 and 0.54, respectively), and it approached an asymptotic value of approximately 95%. In contrast, the spatial variability of throughfall decreased with precipitation amount and I₃₀ in power functions (R² =0.71 and 0.60, respectively), and the steady value of the coefficient of spatial variability was about 10%. Under extreme rainfall conditions with precipitation amount greater than 50 mm, rainfall intensity higher than 30 mm/h, and rainfall duration longer than 12 h, the throughfall percentage and spatial variability of throughfall were still at the steady stage. Canopy thickness and branch angle were the main vegetation factors influencing throughfall. Throughfall percentage was negatively and positively correlated with canopy thickness and branch angle, respectively (P < 0.05). There was an aggregation effect of throughfall for Caragana korshinskii, with an average occurrence frequency of 8.53%. The occurrence of aggregation effects increased with precipitation amount, I₃₀, and rainfall duration, and the aggregation effect was more evident under rainfall conditions with amount larger than 20 mm, intensity higher than 10 mm/h, and duration longer than 10 h. The above results provided detailed information on the characteristics and main influential factors of throughfall, which is important for detecting the dynamic processes of rainfall redistribution and soil moisture for shrubs.