Abstract Especially given the falling cost of hardware and ubiquity of local-area wireless networking technologies, there is high current interest in programming models and software infrastructures to suppor t ubiquitous and environmental computing. In this paper, we argue from both a priori reasoning and our experimental experience that, with slight modifications, the tuplespace programming model is a natural fit for such an application. From the functionality perspective, the constraints of the problem lead directly to those put forth when the tuplespaces were originally proposed. From the systems engineering perspective, the use of a tuplespace enables improved robustness and and more resilient resource management, which are necessary if ubiquitous computing is to achieve high penetration. In the context of our implemented prototype environment, the iRoom interactive workspace, we describe the principles that led us to this design choice, the modifications we made to the basic tuplespace model to improve its suitability for ubiquitous computing, our day-to-day experience over the past year and a half putting these principles into practice, and some avenues for future research. The experience and applications described run on top of infrastructure software tools that we distributing to enable other researchers to build on our efforts.