Activity-based offices are increasingly popular. However, productivity and well-being in these work environments have been little researched. The aims of this study were to quantitatively determine perception and use of the activity-based office environment in relation to self-rated productivity and well-being at work, and to identify important predictors of these outcomes. Four activity-based offices in a large Swedish government agency were surveyed 12 months after implementation.Two hundred and thirty-nine respondents were included in the analyses. Linear regression models, adjusted for relevant covariates, were constructed separately for predictors measuring satisfaction with different aspect of the environment (physical environment, privacy, communication, personalization, personal storage, IT functions and cleaning) and office use (the number of daily workspace switches, different workspaces used and the time spent looking for a workspace). Satisfaction with the physical environment, privacy and communication had the strongest positive association with self-rated productivity and well-being at work. Increased workspace switching was associated with higher productivity, while an increase in self-reported time spent searching for a workspace was associated with lower productivity and well-being. However, predictors related to office use generally explained only a small proportion of variance in the two outcomes. The results suggest that office developers should focus particularly on privacy needs but also on communication, personalization, smooth workspace switching and minimization of work time spent looking for available workspaces.