Fine-needle biopsy (FNB) is associated with problems, such as tumor seeding, which are probably underestimated. The aim of this study was to validate prospectively the accuracy of our diagnostic work-up without FNB, for defining indications for surgery in a cohort of patients with focal liver lesions (FLLs). Between January 1997 and December 1998, 160 consecutive patients carrying 225 FLLs admitted to our department were evaluated prospectively. Preoperative diagnoses were established by means of clinical histories, serum tumor marker levels, ultrasonography, and spiral computed tomography (CT). Angiography, magnetic resonance imaging (MRI), and Lipiodol-CT were performed when it was considered necessary to plan the surgical strategy. All the patients underwent surgery and results of pathological examinations were obtained for all of them. The preoperative diagnoses of 221 of the 225 lesions (98.2%) were confirmed, and the indications for liver resection in 156 of the 160 patients (97.5%) were correct. The respective accuracy, sensitivity, specificity, and positive and negative predictive values were 99.6%, 100%, 98.9%, 99.3%, and 100% for diagnosis of hepatocellular carcinoma (HCC); 99.1%, 100%, 98.8%, 96.9%, and 100% for metastases; 99.6%, 100%, 99.5%, 91%, and 100% for cholangiocellular carcinomas (CCCs); all 100% for mixed HCC-CCCs; and 98.7%, 57.1%, 100%, 100%, and 98.6% for benign tumors. In view of these results, the fact that the real risks of FNB have yet to be established and the possibility that tumor seeding has a major impact on patient prognosis, the use of FNB should be drastically limited.