$BACKGROUND$: Few studies have explored the impact of environmental change on walking using controlled comparisons. Even fewer have examined whose behaviour changes and how. In a natural experimental study of new walking and cycling infrastructure, we explored changes in walking, identified groups who changed in similar ways and assessed whether exposure to the infrastructure was associated with trajectories of walking.

$METHODS$: 1257 adults completed annual surveys assessing walking, sociodemographic and health characteristics and use of the infrastructure (2010-2012). Residential proximity to the new routes was assessed objectively. We used latent growth curve models to assess change in total walking, walking for recreation and for transport, used simple descriptive analysis and latent class analysis (LCA) to identify groups who changed in similar ways and examined factors associated with group membership using multinomial regression.

$RESULTS$: LCA identified five trajectories, characterised by consistently low levels; consistently high levels; decreases; short-lived increases; and sustained increases. Those with lower levels of education and lower incomes were more likely to show both short-lived and sustained increases in walking for transport. However, those with lower levels of education were less likely to take up walking. Proximity to the intervention was associated with both uptake of and short-lived increases in walking for transport.

$CONCLUSIONS$: Environmental improvement encouraged the less active to take up walking for transport, as well as encouraging those who were already active to walk more. Further research should disentangle the role of socioeconomic characteristics in determining use of new environments and changes in walking.