Sorafenib is the first-line drug used for patients with advanced hepatocellular carcinoma (HCC). However, acquired sorafenib resistance in cancer patients limits its efficacy. Here, we performed the first genome-wide CRISPR/Cas9-based screening on sorafenib-treated HCC cells to identify essential genes for non-mutational mechanisms related to acquired sorafenib resistance and/or sensitivity in HCC cells. KEAP1 was identified as the top candidate gene by Model-based Analysis of Genome-wide CRISPR/Cas9 Knockout (MAGeCK). KEAP1 disrupted HCC cells were less sensitive than wild-type cells in short- and long-term sorafenib treatments. Compared to wild-type cells, KEAP1-disrupted cells showed lower basal and sorafenib-induced reactive oxygen species (ROS) levels and were more resistant to oxidative stress-induced cell death. The absence of KEAP1 led to increased activity of Nrf2, a key transcription factor controlling antioxidant responses, as further evidenced by increased expression of Nrf2-controlled genes including NQO1, GPX2 and TXNRD1, which were positively associated with chemoresistance. In addition, KEAP1 disruption counteracted the reduction of cell viability and the elevation of ROS caused by lenvatinib, a drug that recently showed clinical efficacy as a first-line treatment for unresectable HCC. Finally, Keap1 disruption also increased the resistance of cells to regorafenib, a recently approved drug to treat HCC as a second line therapy. Taken together, our data indicate that deregulation of the KEAP1/Nrf2 pathway following KEAP1 inactivation contributes to sorafenib, lenvatinib, and regorafenib resistance in human HCC cells through up-regulation of Nrf2 downstream genes and decreased ROS levels.