To explore the mechanism of surface runoff and sediment yield of evergreen broad-leaved forest in the middle subtropical region, the natural rainfall and raindrop characteristics from May to September in 2016 and 2017 were observed by field observation and indoor test analysis Also, the leading environmental factors were selected by RDA sequencing and the Forward analysis of CANOCO. The results showed that:(1) Sediment yield was positively correlated with rainfall outside forest and throughfall in forest area. When runoff was greater than 0.064 mm in the study area, sediment occurred. The buffer effect of canopy layer on rainfall kinetic energy was more obvious when rainfall intensity was small. (2) Canopy buffering energy, leaf area index and stemflow and sediment yield had a negative correlation, rainfall duration, rainfall outside the forest, forest rainfall, average rainfall intensity, raindrop number, raindrop diameter, raindrop velocity, rainfall kinetic energy and rainfall kinetic energy on runoff sediment produced positive influence. (3) Through the screening of the leading factors of 12 environmental factors, we got the most significant influences of total rainfall, total rainfall, average rainfall intensity, canopy buffer kinetic energy and rainfall duration on the runoff and sediment yield of 12 environmental factors. The results from canonical eigenvalues analysis showed that choosing only 5 main environmental variables could explain 73.2% of the information content of the relationship between environmental factors and runoff and sediment yield, which was only 16.8% less than that explained by 12 variables.