High tibial osteotomy (HTO) and knee joint distraction (KJD) are joint-preserving treatments that unload the more affected compartment (MAC) in knee osteoarthritis. This post-hoc study compares two-year cartilage-thickness changes after treatment with KJD vs. HTO, and identifies factors predicting cartilage restoration. Patients indicated for HTO were randomized to KJD (KJDHTO) or HTO treatment. Patients indicated for total knee arthroplasty received KJD (KJDTKA). Outcomes were the MRI mean MAC cartilage thickness and percentage of denuded bone area (dABp) change two years after treatment, using radiographic joint space width (JSW) as the reference. Cohen’s d was used for between-group effect sizes. Post-treatment, KJDHTO patients (n = 18) did not show significant changes. HTO patients (n = 33) displayed a decrease in MAC cartilage thickness and an increase in dABp, but an increase in JSW. KJDTKA (n = 18) showed an increase in MAC cartilage thickness and JSW, and a decrease in dABp. Osteoarthritis severity was the strongest predictor of cartilage restoration. Kellgren–Lawrence grade ≥3 showed significant restoration (p < 0.01) after KJD; grade ≤2 did not. Effect sizes between severe KJD and HTO patients were large for MAC MRI cartilage thickness (d = 1.09; p = 0.005) and dABp (d = 1.13; p = 0.003), but not radiographic JSW (d = 0.28; p = 0.521). This suggests that in knee osteoarthritis patients with high disease severity, KJD may be more efficient in restoring cartilage thickness.