Government agencies often face trade-offs in developing initiatives that address a public good given competing concerns of various constituent groups. Efforts to construct data warehouses that enable data mining of citizens’ personal information obtained from other organizations (including sister agencies) create a complex challenge, since privacy concerns may vary across constituent groups whose priorities diverge from agencies’ e-government goals. In addition to privacy concerns, participating government agencies’ priorities related to the use of the information may also be in conflict. This paper reports on a case study of the Integrated Non-Filer Compliance system used by the California Franchise Tax Board for which data are collected from federal, state and municipal agencies and other organizations in a data mining application which aims to identify residents who under-report income or fail to file tax returns. This system pitted the public good (ensuring owed taxes are paid) with citizen concerns about privacy. Drawing on stakeholder theory, we propose a typology of four stakeholder groups (data controllers, data subjects, data providers and secondary stakeholders) to address privacy concerns and argue that by ensuring procedural fairness for the data subjects, agencies can reduce some barriers that impede the successful adoption of e-government applications and policies. The paper concludes that data controllers can reduce adoption and implementation barriers when e-government data mining applications rely on data shared across organizational boundaries: identify legitimate stakeholders and their concerns prior to implementation; enact procedures to ensure procedural fairness when data are captured, shared, and used, explain to each constituency how the data mining application helps to ensure distributive fairness, and continue to gauge stakeholders’ responses and ongoing concerns as long as the application is in use.