Accurate space-use prediction helps architects to optimize space efficiency in buildings, thereby achieving economic and environmental sustainability. However, current space-use prediction models and approaches either disregard or oversimplify the role of building users' space preferences in spatial-choice behaviour, thereby compromising prediction accuracy. The aim of this study was thus to develop a space-preference model of spatial-choice behaviour with a focus on group work-related activities. A total of 2464 observations of spatial choices were collected using a discrete-choice experiment. The data were modelled using a conditional logit model and then validated in a predictive success test. The resulting model clearly explains space preferences for group work-related activities and predicts spatial-choice behaviour by generating space-use probabilities for given spaces. The model is compared to a space-preference model for individual work-related activities. Lastly, the application of the model was demonstrated in a case example.