This study examined whether supervised peer mentoring was associated with adaptation trajectories among first-year repatriate students and explored programme-record patterns that could help interpret the observed quantitative trajectories. A longitudinal quasi-experimental mixed-methods design was used with 400 students: 204 in the mentoring condition and 196 in a delayed-access comparison condition. Academic adaptation, perceived social support, and acculturative stress were assessed at three time points during the 2024–2025 academic year. Quantitative data were analysed using three-level linear mixed-effects models with academic-group random intercepts, covariate adjustment, and CR2 cluster-robust sensitivity analyses. By T3, the mentoring condition showed more favourable changes in academic adaptation (Hedges’ g = 0.39), perceived social support (g = 0.29), and acculturative stress (g = −0.35). Qualitative analysis of 222 mentor reports, supervision records, and de-identified case summaries identified recurrent documentary patterns involving procedural clarification, lower-threshold help-seeking, and supported social participation. It also identified low-engagement cases, unresolved institutional barriers, mentor-dependence risks, and procedural gains without broader social integration. These findings support supervised mentoring as a feasible adjunct to usual university services. Because allocation was non-random and institutional translation was not directly measured, the results should be interpreted as programme-associated differences and theoretically informed process interpretations rather than as causal or mediation evidence.