『Abstract
　We used stream chemistry and hydrogeomorphology data from 549 stream and 447 river sites to estimate NO3-N removal in the Upper Mississippi, Missouri, and Ohio Rivers. We used two N removal models to predict NO3-N input and removal. NO3-N input ranged from 0.01 to 338 kg km^-1 d^-1 in the Upper Mississippi River to ＜0.01-54 kg km^-1 d^-1 in the Missouri River. Cumulative river network NO3-N input was 98700-1012676 Mg year^-1 in the Ohio River, 85961-89288 Mg year^-1 in the Upper Mississippi River, and 59463-61541 Mg year^-1 in the Missouri River. NO3-N output was highest in the Upper Mississippi River (0.01-329 kg km^-1 d^-1), followed by the Ohio and Missouri Rivers (＜0.01-236 kg km^-1 d^-1) sub-basins. Cumulative river network NO3-N output was 97499 Mg year^-1 for the Ohio River, 84361 Mg year^-1 for the Upper Mississippi River, and 59200 Mg year^-1 for the Missouri River. Proportional NO3-N removal (PNR) based on the two models ranged from ＜0.01 to 0.28. NO3-N removal was inversely correlated with stream order, and ranged from ＜0.04 to 8.57 kg km^-1 d^-1 in the Upper Mississippi River to ＜0.001-1.43 kg km^-1 d^-1 in the Missouri River. Cumulative river network NO3-N removal predicted by the two models was: Upper Mississippi River 4152 and 4152 Mg year^-1, Ohio River 3743 and 378 Mg year^-1, and Missouri River 2277 and 197 Mg year^-1. PNR removal was negatively correlated with both stream order ( r = -0.80--.87) and the percent of the catchment in agriculture (r = -0.38-0.76).

Keywords: Nitrogen removal; River networks; Mississippi River basin』

Introduction
Methods
　Site selection
　Stream chemistry
　River networks, hydrology, and land use
　Nitrate input, output and removal
　Statistical analyses
Results
Discussion
Acknowledgements
References