Highly promising preclinical data obtained in cultured cells and in nude mice bearing xenografts contrast with the rather modest clinical efficacy of Polo-like kinase 1 (Plk1) inhibitors. In the present study, we investigated if Plk1 might be a suitable target in hepatocellular carcinoma (HCC) and if a genetically engineered mouse tumor model that well reflects the tumor cell and micro-environmental features of naturally occurring cancers might be suitable to study anti-Plk1 therapy. Analysis of Plk1 expression in human HCC samples confirmed that HCC express much higher Plk1 levels than the adjacent normal liver tissue. Inhibition of Plk1 by an adenovirus encoding for a short hairpin RNA against Plk1 or by the small-molecule inhibitor BI 2536 reduced the viability of HCC cell lines and inhibited HCC xenograft progression in nude mice. Treatment of transforming growth factor (TGF) α/c-myc bitransgenic mice with BI 2536 during hepatocarcinogenesis reduced the number of dysplastic foci and of Ki-67-positive cells within the foci, indicating diminished tumorigenesis. In contrast, BI 2536 had no significant effect on HCC progression in the transgenic mouse HCC model as revealed by magnetic resonance imaging. Measurement of BI 2536 by mass spectrometry revealed considerably lower BI 2536 levels in HCC compared with the adjacent normal liver tissue. In conclusion, low intratumoral levels are a novel mechanism of resistance to the Plk1 inhibitor BI 2536. Plk1 inhibitors achieving sufficient intratumoral levels are highly promising in HCC treatment.