Background Contact with natural environments can bring health benefits, but research is lacking on how changes in access to natural environments might improve health, especially for deprived populations. Objective To evaluate the health impacts of woodland environment interventions intended to increase communities’ engagement with these woodlands. Design A prospective study of Forestry Commission Scotland’s Woods In and Around Towns (WIAT) programme in deprived communities to enhance public access to natural environments. The study investigated the impact that WIAT had on community-level mental health over time. Setting Three intervention and three control woodland sites, and associated communities within 1.5 km of the woodlands, located in central Scotland and eligible for WIAT support. Participants A core community survey was administered at each site in three waves, at baseline and after each phase of intervention (n = 5460, panel A). The completed survey contained a nested longitudinal cohort (n = 609, panel B). Community members also undertook 6-monthly environmental audits at all sites (n = 256) and participated in post-intervention focus groups (n = 34). Interventions Phase 1 involved physical changes to the woodlands, including footpaths, entrances and vegetation. Phase 2 involved community engagement events promoting woodland use. Main outcome measures The primary outcome was the Perceived Stress Scale (PSS). Other health measures included health-related quality of life (HRQoL) EuroQol-5 Dimensions (EQ-5D), physical activity (PA) [International Physical Activity Questionnaire (IPAQ)], connectedness to nature [Inclusion of Nature in Self (INS) scale] and social cohesion. Results The PSS scores significantly increased in the intervention group and marginally decreased in the control group. Multilevel regression models showed a differential impact between the intervention and the control at survey wave 3 in panel A [B(unstandardised coefficient) 3.58, 95% confidence interval (CI) 2.85 to 4.31;p < 0.001] and in panel B [B3.03, 95% CI 1.54 to 4.52;p < 0.001]. Using the same analytical approach, no significant change in HRQoL was associated with the intervention. Economic assessment included an illustrative cost–utility analysis and a cost–consequences analysis. The differential in stress between the intervention group and the control group was lower or non-significant in those who visited ‘nature’ in the previous year [panel A,B1.9, 95% CI 0.8 to 3.0;p < 0.001; panel B,B0.64, 95% CI –1.60 to 2.88;p = 0.57]. The IPAQ score showed a positive association with the intervention for moderate levels of PA [panel B,B559.3, 95% CI 211.3 to 907.2;p = 0.002] and overall PA [panel B,B861.5, 95% CI 106.5 to 1616.4;p = 0.025]. The intervention was also associated with increased nature connectedness and social cohesion by wave 3 – significant for panel A only. Qualitative and quantitative evidence showed that interventions increased the perceived quality of the woodland environment and enhanced its enjoyment for different activities, but the increase in use of natural environments post intervention was only 6% (panel B). Limitations This study was limited to three intervention sites. External factors may be the primary influence on health outcomes. Conclusions The WIAT interventions did not improve community-level health within 6 months of completion, and hence there was no basis for demonstrating cost-effectiveness. However, the WIAT interventions are low cost (average £11.80 per person in the eligible population) and have potential for cost-effectiveness, if health benefits were found in the longer term. Future work Using routinely collected data to consider a whole-programme evaluation is recommended. Funding The National Institute for Health Research Public Health Research programme.