The objective of this research was to experimentally test the effect of tillage methods on the reduction of runoff, non-point source (NPS) pollution load, sediment and discharge under a rainfall simulation. We used the runoff plots of $5m{ imes}30m$ ($L{ imes}W$) in size. Experimental treatments were conventional tillage (CT), CT-rice straw bundle (CT-RSB) and two no-till (NT) plots; slope of 3 % or 8 % ; and rainfall intensity of 30 mm/hr. The rainfall simulation was conducted to three times. The time to initial runoff from NT plots was less than that from CT plots regardless of the slope and it was delayed about 65~90 % compared to that of CT plot. And sediment discharge of 8 % slope reduced to 55 % compared to CT plot. But the sediment discharge was not occurred at 3 % slope. The NT and CT-RSB methods have a great possibility of reducing runoff and NPS pollution loads. Runoff rate of NT plots was significantly lower than those of CT plot. The average NPS pollution loads of the NT plots and CT-RSB plot reduced about 55~80 % and 2.1~40 % compared to those of the CT plots, respectively. It was also shown that runoff and NPS pollution loads reduction by NT method could be very significant and contribute to improve the water quality of streams in agricultural regions. It was concluded that the use of NT method on agricultural fields could reduce soil erosion and muddy runoff significantly and help improve the water quality and aquatic ecosystem.