Neighborhood walkability shows significant positive relationship with residents' walking for transport in cross-sectional studies. We examined prospective relationships of neighborhood walkability with the change in walking behaviors over 4 years among middle-to-older aged adults (50-65 years) residing in Adelaide, Australia. A baseline survey was conducted during 2003-2004, and a follow-up survey during 2007-2008. Walking for transport and walking for recreation were assessed at both times among 504 adults aged 50-65 years living in objectively determined high- and low-walkable neighborhoods. Multilevel linear regression analyses examined the associations of neighborhood walkability with changes over 4 years in walking for transport and walking for recreation. On average, participants decreased their time spent in walking for transport (-4.1 min/day) and for recreation (-3.7 min/day) between the baseline and 4-year follow-up. However, those living in high-walkable neighborhoods showed significantly smaller reduction (adjusted mean change: -1.1 min/day) in their time spent in walking for transport than did those living in low-walkable neighborhoods (-6.7 min/day). No such statistically-significant differences were found with the changes in walking for recreation. High-walkable neighborhoods may help middle-to-older aged adults to maintain their walking for transport.