To connect aberrant phospho-signaling events to therapeutic vulnerabilities in hepatocellular carcinoma (HCC), we performed a mass-spectrometry-based phosphoproteomic quantification of human HCC tumors, adjacent tissues and cell lines. We developed an integrated pipeline as the inference of druggable kinome (iDruKin), and computationally inferred differentially activated protein kinases from the phosphoproteomic data, and totally identified 16 kinases as potent drug targets in HCC. Together with experiments in cell lines, tissues and mice, our results not only confirmed MTOR as a major target in HCC, but also predicted the inhibition of CDK4/6 and the activation of PKCs and AMPKs to be promisingly effective in HCC therapy. Additionally, we observed most of these kinases are more significantly dysregulated in recurrent HCCs, and the kinase-activity signature was used to accurately classify HCC samples. Thus, this study demonstrated the potential of iDruKin in identifying drug targets and prognostic markers for HCC and other cancers.