Background: In parallel with a multicenter randomized controlled trial that reported an equal recurrence-free survival (RFS) of early-stage hepatocellular carcinoma (HCC) patients who underwent either surgery (SUR) or radiofrequency ablation (RFA), we also enrolled HCC patients who fulfilled the enrollment criteria but did not give consent to participate in the RCT. Methods: All patients gave informed consent to participate in this study. Inclusion criteria were as follows: primary HCC with less than or equal to 3 tumors, each measuring 3 cm or smaller; without vascular invasion or extrahepatic metastasis; Child-Pugh score of 7 or less; and ages between 20 and 79 years. The feasibility for both treatments was confirmed by a joint chart review by surgeons and hepatologists. The primary endpoint was RFS and overall survival. A pre-specified interim analysis was performed to compare RFS. Results: Between April 2009 and August 2015, 740 patients (371 in SUR, 369 in RFA) were enrolled from 49 participating hospitals in Japan. The SUR group had significantly fewer patients with chronic hepatitis C (56.6% vs. 69.4%), higher median value of platelet count (145 vs. 120 × 10⁹/L), and more patients with > 2 cm tumors (49.9% vs. 27.9%); most patients had a single tumor (91.1% vs. 88.3%). During the median follow-up period of 5 years, tumor recurrence was observed in 192 of SUR and 218 of RFA with 3-year RFS being 66.0% and 61.7%, respectively (P = 0.091). In subgroup analysis, RFS was significantly better in SUR in patients with ≤ 2 cm tumors (62.9% vs. 51.7% in 3 years; hazard ratio [HR] 0.72, 95% confidence interval [CI] 0.56-0.93; P = 0.014), whereas the difference was not significant in those with > 2 cm tumors (52.7% vs. 46.4%; HR 0.85, 95% CI 0.63-1.18; P = 0.34). The adjusted HR for RFS using inversed probability of treatment weighting was 0.89 (95% CI, 0.72-1.10; P = 0.287). Conclusions: The imbalance in patient characteristics reflected a real-world practice. Factors related to background liver disease rather than tumor characteristics might have a larger impact on the recurrence in early HCC. Clinical trial information: C000001796.