Various analytical items of surface water, subsurface water, irrigation water and rain water were measured in experimental plots at the university farm. Three experimental plots were set up: 1. A conventional plot, where ordinary plowing and puddling were practiced with the application of conventional doses of N-P-K as the basal and two top-dressings. 2. No-puddling plot, where plowing, but no puddling, was practiced before transplanting and all the N was applied to the nursery box in the form of a coated fertilizer (containing 30% N and 10% K_2O), with P and K supplemented as the basal before plowing. 3. No-fertilizer plot, where plowing and puddling were practiced but no fertilizers nor pesticides were applied. The surface effluent loads of the no-puddling plot were 31% for T-N, 25% for T-P, 55% for T-COD, 94% for BOD and 27% for SS as compared with those of the conventional plot. The subsurface effluent loads of the no-puddling plot were, likewise, 52% for D-N, 87% for D-P, 78% for D-COD and 89% for BOD. The effluent loads (N, P, COD, BOD, SS) of the conventional plot were greater than the incoming loads, where those of the no-puddling plot were greater for P, COD and BOD but smaller for N and SS than the incoming loads. Thus, the nursery box application of coated fertilizer in combination with no-puddling appeared to be effective in reducing the effluent N and SS loads from paddy fields.