Proton beam radiotherapy (PBT) has recently been applied to treat hepatocellular carcinoma (HCC); however, there is no randomized controlled trial-based evidence on its safety and efficacy. We compared the outcomes of PBT and radiofrequency ablation (RFA) in patients with recurrent/residual HCC (rHCC) in a phase III non-inferiority trial.Patients with rHCC (size <3 cm, number ≤2) were randomly assigned to receive PBT or RFA according to the Child-Pugh score and tumor stage. After randomization, if the assigned treatment was technically infeasible, crossover was allowed. The primary endpoint was 2-year local progression-free survival (LPFS), with a non-inferiority margin of 15% in per-protocol (PP) population and complementarily in intention-to-treat (ITT) population (NCT01963429).The ITT population comprised 144 PBT (n=72) and RFA (n=72) patients. Six patients in PBT arm switched to RFA, while 19 in RFA arm switched to PBT. In the PP population, the 2-year LPFS rate with PBT (n=80) vs. RFA (n=56) was 94.8% vs. 83.9%, a difference of 10.9 percentage points (90% confidence interval [CI], 1.8-20.0; p < 0.001); in the ITT population, the 2-year LPFS rate with PBT vs. RFA was 92.8% vs. 83.2%, a difference of 9.6 percentage points (90% CI, 0.7-18.4; p <0.001), meeting the criteria for non-inferiority. The 3- and 4-year LPFS rates for PBT were also non-inferior to those for RFA. The most common adverse events (AEs) were radiation pneumonitis (32.5%) and decreased leukocyte counts (23.8%) for PBT and increased alanine aminotransferase levels (96.4%) and abdominal pain (30.4%) for RFA. No Grade 4 AEs or mortality were noted.PBT showed LPFS values that were non-inferior to those for RFA and were tolerable and safe.