The goal of this study was to determine whether a panel of tumor suppressor gene markers of allelic loss could serve as a representative indicator of gene damage and thereby provide further discriminative power over current staging systems for recurrence-free prognostication in patients undergoing liver transplantation in the presence of hepatocellular carcinoma. The paraffin blocks from 103 cases of hepatocellular carcinoma were obtained, and cellular targets were selected for tissue microdissection genotyping. Tumor suppressor gene loss was based on loss of heterozygosity situated within or adjacent to specific genes of interest (APC, CDKN2A, DCC, MET, MYC1, OGG1, p34, p53, PTEN). Microdissected tissue was amplified using polymerase chain reaction (PCR) with flanking oligonucleotides bearing fluorescent labels designed for GeneScan fragment analysis; PCR products were separated by capillary electrophoresis. Normal microdissected tissue samples for each case were evaluated for informative status with respect to individual alleles for 18 microsatellites at 10 genomic loci-1p, 3p, 5q, 7q, 8q, 9p, 10q, 17p, 17q, 18q. The measure of allelic loss of heterozygosity combined with tumor number, tumor size, vascular invasion, lobar distribution, and patient gender provide a highly discriminatory model for predicting cancer recurrence after liver transplantation. Using our previously developed artificial neural network model in combination with the genotyping results, unambiguous predictions were made for 91 of the103 patients (88.3%).Of these, 1 was lost to follow-up, and 9 died recurrencefree less than 3 years posttransplantation. For the remaining 81, the combined models predicted tumor recurrence outcomes with complete accuracy. Microdissection genotyping provides powerful supplementary discriminative information for tumor-free survival. (Liver Transpl 2003; 9:664-671.)