Cognitive and physical activities can benefit cognition. However, knowledge about the neurobiological mechanisms underlying these activity-induced cognitive benefits is still limited, especially with regard to the role of white matter integrity (WMI), which is affected in cognitive aging and Alzheimer’s disease. To address this knowledge gap, we investigated the immediate and long-term effects of cognitive or physical training on WMI, as well as the association between cognitive and physical lifestyles and changes in WMI over a 6-month period. Additionally, we explored whether changes in WMI underlie activity-related cognitive changes, and estimated the potential of both trainings to improve WMI by correlating training outcomes with WMI. In an observational and interventional pretest, posttest, 3-month follow-up design, we assigned 47 community-dwelling older adults at risk of dementia to 50 sessions of auditory processing and working memory training (n = 13), 50 sessions of cardiovascular, strength, coordination, balance and flexibility exercises (n = 14), or a control group (n = 20). We measured lifestyles trough self-reports, cognitive training skills through training performance, functional physical fitness through the Senior Fitness Test, and global cognition through a cognitive test battery. WMI was assessed via a composite score of diffusion tensor imaging-based fractional anisotropy (FA) of three regions of interest shown to be affected in aging and Alzheimer’s disease: the genu of corpus callosum, the fornix, and the hippocampal cingulum. Effects for training interventions on FA outcomes, as well as associations between lifestyles and changes in FA outcomes were not significant. Additional analyses did show associations between cognitive lifestyle and global cognitive changes at the posttest and the 3-month follow-up (β ≥ 0.40, p ≤ 0.02) and accounting for changes in WMI did not affect these relationships. The targeted training outcomes were related to FA scores at baseline (cognitive training skills and FA composite score, rs = 0.68, p = 0.05; functional physical fitness and fornix FA, r = 0.35, p = 0.03). Overall, we found no evidence of a link between short-term physical or cognitive activities and WMI changes, despite activity-related cognitive changes in older adults at risk of dementia. However, we found positive associations between the two targeted training outcomes and WMI, hinting at a potential of long-term activities to affect WMI.