Background Physical activity interventions have had varying results on modifying hippocampal volume. Methods The Retirement in Action (REACT) study conducted a randomised-controlled trial of a 12-month physical activity and behaviour maintenance intervention in older adults at risk of mobility impairments. The physical activity sessions were delivered twice weekly for the first twelve weeks, and then reduced to once weekly, to groups of 15 participants. Activities included cardiovascular, strength, balance and flexibility exercises. A sub-sample of participants in the physical activity (N = 54) and control arms (N = 48) underwent a 3T MRI brain scan and cognitive assessments at baseline, 6- and 12-months (mean age = 76.6 years, 6.8 SD). It was hypothesised that the intervention would lead to a reduced rate of decline in hippocampal volume. Group differences in changes in cognition were also examined. Results As hypothesised, we found a maintenance in left hippocampal volume in the intervention arm, in comparison with the control arm after 12 months (p=0.027). In a secondary analysis, this effect was attenuated after including age, sex and education level as covariates (p = 0.057). There was no significant between-group difference in the right hippocampus (p = 0.405). Contrary to our hypothesis, we did not find a beneficial effect of the intervention on cognitive outcomes. Conclusions Our findings suggest that a community-based physical activity intervention can significantly ward-off hippocampal atrophy in older adults. While the lack of effects on cognition may limit the interpretability of our results, our findings of hippocampal maintenance are promising given the potential clinical relevance of protecting the hippocampus from age-related decline.