Many older cities in the United States confront the problem of long-term declines in population and economic activity in certain neighborhoods have resulted in blighted conditions that make conventional revitalization initiatives based on increased residential and commercial development unlikely to succeed. Planning scholars have developed a theory of smart shrinkage in which emphasis is placed on non-residential land uses that can maintain and improve quality of life while positioning some land for future growth-oriented activities (Hollander and Németh 2011). Smart shrinkage research and practice involves application of methods from information technology and decision science to identify vacant and abandoned parcels for acquisition and redevelopment for alternative uses in such a way as to meet multiple social and economic goals while respecting myriad resource constraints (Johnson, Hollander and Hallulli 2013; Johnson 2011). This paper addresses the following questions: How do planning practitioners conceptualize the problem of smart shrinkage? In what ways do the decision processes they apply in practice diverge from the theory of decision modeling? How can planners make most appropriate use of information and decision sciences to develop sustainable and politically-feasible strategies for smart decline? To answer these questions we describe a project with planners from Baltimore, Maryland (Johnson and Hollander 2014; Davenport Whiteman 2014) that uses spatial data and spatial and decision analytic methods to select aggregations of parcels, called clusters, for alternative land uses that jointly optimize multiple objectives. These land uses consist of: urban farming, wastewater management and site stabilization for future development. We describe decision modeling results consisting of alternative cluster development strategies, and contrast our model’s recommendations with redevelopment choices actually made by client practitioners. In so doing, we articulate a theory of decision modeling for smart shrinkage that we believe will make best use of data, technology and planner expertise to generate novel and effective strategies for blighted neighborhood stabilization and revitalization. This methodology adapts principles from participatory action research, community information technology and community-based operations research. This study provides a foundation for practitioners to make better use of large volumes of data describing blighted communities, accommodate diverse attitudes about policy and planning responses to blight, and judiciously apply advanced methods in data analysis and decision models. In addition, our study extends theory for urban planners and public sector operations researchers, and provides insights for planning education.