We analyzed the p16INK4 status of 6 hepatocellular carcinoma (HCC) cell lines and 32 primary HCC tumors, including 9 early-stage tumors, to determine whether p16INK4 tumor-suppressor gene inactivation participates in hepatocarcinogenesis. p16INK4 was studied at its protein level through Western blotting, at its messenger RNA (mRNA) level through reverse-transcriptase polymerase chain reaction analysis (RT-PCR) and Northern blotting, and at its genomic level through Southern blotting and PCR-single-strand conformation polymorphism analysis. The p16 protein was absent from 3 of 6 cell lines (50%) and 11 of 32 primary tumors (34%), but present in noncancerous tissues, indicating that p16INK4 is involved in hepatocarcinogenesis. Furthermore, we suggest that the p16 protein loss may contribute to the following: (1) early-stage hepatocarcinogenesis, because it was observed in 22% of early stage tumors; and (2) tumor progression, because it occurred approximately twice as often in advanced rather than in early stage tumors (40%). It was striking that neither p16INK4 homozygous deletion and mutation nor loss of p16INK4 mRNA expression were observed in HCC cell lines and primary tumors, including those specimens from which the p16 protein was absent except the Li7HM cell line, in which p16INK4 mRNA was not detected. These results suggest that p16INK4 in HCC is inactivated predominantly by posttranscriptional regulation rather than by genomic aberrations and lack of transcription.