Surgeons may be severely criticized from the perspective of evidence-based medicine because the majority of surgical publications appear not to be convincing. In the top nine surgical journals in 1996, half of the 175 publications refer to pilot studies lacking a control group, 18% to animal experiments, and only 5% to randomized controlled trials (RCT). There are five levels of clinical evidence: level 1 (randomized controlled trial), level 2 (prospective concurrent cohort study), level 3 (retrospective historical cohort study), level 4 (pre-post study), and level 5 (case report). Recently, a Japanese evidence-based guideline for the surgical treatment of hepatocellular carcinoma (HCC) was made by a committee (Chairman, Professor Makuuchi and five members). We searched the literature using the Medline Dialog System with four keywords: HCC, surgery, English papers, in the last 20 years. A total of 915 publications were identified systematically reviewed. At the first selection (in which surgery-dominant papers were selected), 478 papers survived. In the second selection (clearly concluded papers), 181 papers survived. In the final selection (clinically significant papers), 100 papers survived. The evidence level of the 100 surviving papers is shown here: level-1 papers (13%), level-2 papers (11%), level-3 papers (52%), and level-4 papers (24%); therefore, there were 24% prospective papers and 76% retrospective papers. Here, we present a part of the guideline on the five main surgical issues: indication to operation, operative procedure, peri-operative care, prognostic factor, and post-operative adjuvant therapy.