Background <p>Creating or improving urban parks is a promising intervention for promoting physical activity (PA). However, robust natural experimental evidence of effectiveness remains limited, particularly regarding long-term impacts. We previously showed that co-designed, sustainable urban park improvements in a deprived UK area increased walking and other wellbeing-related behaviours up to 15 months post-intervention. We now examine whether these effects were sustained five years post-intervention (seven years after baseline).</p> Methods <p>Two intervention sites were matched to two comparison sites, with two additional nearby comparison sites included to assess wider neighbourhood trends. Outcomes were assessed using systematic observations at baseline (2018), 15 months post-intervention (2021), and five years post-intervention (2025). The primary outcome was change in the number of people walking; secondary outcomes included vigorous PA, sedentary behaviour, social interactions, and taking notice of the environment. Multilevel mixed-effects negative binomial regression models compared changes between intervention and comparison sites, adjusting for day of week, time of day, and precipitation. Additional analyses compared 15-month (2021) and five-year (2025) follow-ups to assess how intervention effects evolved over time. Intercept surveys assessed self-reported outdoor space use at baseline (2019; <i>n</i> = 217) and follow-up (2025; <i>n</i> = 232).</p> Results <p>From baseline to five years post-intervention, walking at the intervention sites significantly increased by 70% relative to the comparison sites (IRR = 1.70, 95% CI 1.02–2.82). Significant increases were also observed for sedentary behaviour, social interactions, and taking notice of the environment. Comparison of the 15-month and five-year follow-ups indicated some attenuation of intervention effects over time. Observations at nearby comparison sites suggested that these effects were not attributable to wider neighbourhood trends. The largest increases in observed park use were among young people and non-white ethnic groups. Intercept surveys corroborated the observational findings, showing significantly greater increases in self-reported outdoor space use in the intervention area.</p> Conclusions <p>Co-designed, sustainable urban park improvements can generate long-term increases in walking and other wellbeing-related behaviours in deprived urban areas, with effects persisting for at least five years. Urban green space interventions, which can target areas with the greatest need, therefore represent an effective long-term strategy for increasing PA.</p> Study protocol <p>Study protocol published in Open Science Framework before the first follow-up data collection finished (<a href="https://osf.io/zqgcn">https://osf.io/zqgcn</a>). Date of registration: 18 August 2020.</p>