Background To evaluate the provisional long-term durability of group cognitive behavioral therapy (CBT) benefits 8 years post-baseline and to explore tentative associations between digital booster engagement and the maintenance of treatment gains in adults with ADHD. Methods Forty-eight participants recruited from outpatient clinics across Japan (Mage = 39, 87% female) were randomized to group CBT or treatment-as-usual (primary outcome self-reported). At 5 years, CBT participants were offered a digital booster with periodic assessments (optional; non-random uptake). The primary outcome was CAARS Inattention/Memory; secondary outcomes covered ADHD symptoms, anxiety, depression, functioning, and time management. Follow-up occurred at 8 years. Intention-to-treat linear mixed models tested Group × Time effects. Sensitivity analyses included multiple imputation and three-way comparisons (CBT+booster vs. CBT-only vs. treatment-as-usual); exploratory analyses used Bayesian and within-subject models. Results At 8-year follow-up, 24 participants (50%) completed assessments. CBT nominally showed lower inattention than TAU (p = .038, d = -0.65), though results were not robust to imputation. No adverse events were recorded. DSM inattentive symptoms (p = .025, ηp² = .09) and family functioning (p = .025, ηp² = .08) appeared more robust and survived imputation. Bayesian models yielded consistent preliminary support. Reliable Change Index analyses indicated higher rates of reliable individual-level improvement in the CBT group (43%) than in the Control group (10%), with all CBT improvers belonging to the booster subgroup. Conclusions This 8-year follow-up (representing a rare longitudinal window) provides preliminary and exploratory evidence that group CBT benefits on inattention and specific functional domains may persist in some adults with ADHD. However, firm conclusions are precluded by the small sample size, high attrition, non-random booster allocation, and reliance on self-reports within a predominantly female Japanese sample. These results should be viewed as hypothesis-generating, requiring replication in larger, more diverse cohorts.  Trial registration: UMIN000026916 (2017/04/09)