Human activities are rapidly increasing the global supply of reactive N and substantially altering the structure and hydrologic connectivity of managed ecosystems. There is long-standing recognition that N must be removed along hydrologic flowpaths from uplands to streams, yet it has proven difficult to assess the generality of this removal across ecosystem types, and whether these patterns are influenced by land-use change. To assess how well upland nitrate (NO3-) loss is reflected in stream export, we gathered information from >50 watershed biogeochemical studies that reported nitrate concentrations ([NO3-]) for stream water and for either upslope soil solution or groundwater NO3- to examine whether stream export of NO3- accurately reflects upland NO3- losses. In this dataset, soil solution and streamwater [NO3-] were correlated across 40 undisturbed forest watersheds, with streamwater [NO3-] typically half (median = 50%) soil solution [NO3-]. A similar relationship was seen in 10 disturbed forest watersheds. However, for 12 watersheds with significant agricultural or urban development, the intercept and slope were both significantly higher than the relationship seen in forest watersheds. Differences in concentration between soil solution or groundwater and stream water may be attributed to biological uptake, microbial processes including denitrification, and/or preferential flow routing. The results of this synthesis are consistent with the hypotheses that undisturbed watersheds have a significant capacity to remove nitrate after it passes below the rooting zone and that land use changes tend to alter the efficiency or the length of watershed flowpaths, leading to reductions in nitrate removal and increased stream nitrate concentrations.