Technical Abstract: Comprehensive, long-term data for watershed systems across diverse locations are essential for interdisciplinary hydrologic and ecosystem analysis and model development, calibration and validation. The USDA and Agricultural Research Service (ARS) have supported watershed research since the 1930¿s with sites added periodically to meet evolving needs. However data from ARS watersheds have been managed and disseminated independently at each research location, hindering accessibility and utility of these data for policy-relevant, multi-site analyses. A team within the Conservation Effects Assessment Project - Watershed Assessment Studies has developed a web-based data delivery system to provide access to soil, water, climate, land-management, and socio-economic data from fourteen watersheds. The system, STEWARDS: Sustaining the Earth¿s Watersheds, Agricultural Research Data System, allows a variety of users to search, visualize, and download data via the internet. STEWARDS consists of: 1) a centralized site with Web/SQL/ArcGIS servers and application software, including a database management system (DBMS) and a geospatial data access portal; 2) data: including measurement data, imagery/GIS, and metadata; 3) users; and 4) research watershed sites that are data sources. Development of STEWARDS presented many challenges that were met to deliver the data system on time, according to requirements, and within available resources. Managing a complex virtual team that spanned diverse organizational units required communication and trust. It was essential that IT specialists understand that vague and changing requirements are reasonable for a system to support loosely coupled research across diverse long-term watershed projects. Researchers and data managers had to learn how to clearly communicate about their data to persons who are not experts in their discipline and yet retain adequate detail to meet the requirements of disciplinary specialists. Open communication, respect for the perspectives and constraints of other participants, and a shared commitment to the mission of the team provided the basis for this trust. Approaches used by the development team, both in project management and in system design, may provide a model for other scientific data management applications. Anticipated benefits of STEWARDS include preservation of data, increased data use, and facilitation of hydrological research within and across watersheds with diverse collaborators.