Although the watershed approach has long been used to study whole-ecosystem function, it has seldom been applied to study human-dominated systems, especially those dominated by urban and suburban land uses. Here we present 3 years of data on nitrogen (N) losses from one completely forested, one agricultural, and six urban/suburban watersheds, and input-output N budgets for suburban, forested, and agricultural watersheds. The work is a product of the Baltimore Ecosystem Study, a long-term study of urban and suburban ecosystems, and a component of the US National Science Foundation's long-term ecological research (LTER) network. As expected, urban and suburban watersheds had much higher N losses than did the completely forested watershed, with N yields ranging from 2.9 to 7.9 kg N ha-1 y-1 in the urban and suburban watersheds compared with less than 1 kg N ha-1 y -1 in the completely forested watershed. Yields from urban and suburban watersheds were lower than those from an agricultural watershed (13-19.8 kg N ha-1 y-1). Retention of N in the suburban watershed was surprisingly high, 75% of inputs, which were dominated by home lawn fertilizer (14.4 kg N ha-1 y-1) and atmospheric deposition (11.2 kg N ha-1 y-1). Detailed analysis of mechanisms of N retention, which must occur in the significant amounts of pervious surface present in urban and suburban watersheds, and which include storage in soils and vegetation and gaseous loss, is clearly warranted.