This paper presents an experiment in which people performed a co-design task in urban design, using a multi-user touch table application with or without interactive model simulations. We hypothesised that using the interactive model simulations would improve communication and co-operation between co-design participants, would help participants to develop shared understanding and would positively affect the co-design process and its outcomes. However, our experiment (involving 60 people in 20 codesign sessions) only partly confirmed these hypotheses. People positively evaluated the interactive model simulation tools (an interactive map of an urban area, interactive models for traffic, sound, sight and safety, and ‘tangibles’), and these tools promoted communication and co-operation, and the exploration of design solutions. However, people’s experiences of social cohesion and their satisfaction with their own contribution to the co-design process were better without these tools, possibly because using these tools drew people’s attention towards these interactive model simulations and away from the dynamics between the participants. We therefore advocate using such tools selectively, for example, early on in a co-design process, to improve shared understanding of the contents of the problem, rather than later on, when people need to focus on their fellow participants and on the processes of communication and cooperation.