As a safe and efficacious drug, crizotinib was approved by the U.S. Food and Drug Administration (FDA) in 2011 for the treatment of advanced fusion-type nonsmall-cell lung cancer. Although high response ratio was detected from the patients treated with crizotinib, the cancer has eventually conferred resistance to crizotinib. Several drug resistance mutations have been found in the anaplastic lymphoma kinase (ALK) tyrosine kinase domain as the target for crizotinib, but the drug resistance mechanisms remain unclear. Therefore, in this study, the adaptive biasing force (ABF) method and two-end-state free energy calculation approaches were employed to elucidate the resistance mechanisms of crizotinib induced by the mutations L1152R, G1202R, and S1206Y. The ABF simulation results suggest that the reaction coordinates for the unbinding processes of crizotinib from the binding pockets of the mutated ALKs is different from that of the wild type ALK. The potentials of mean force for the crizotinib unbinding and the binding free energies predicted by the two-end-state free energy calculations are consistent with the experimental data. Our results indicate that the three mutations weaken the binding affinity of crizotinib obviously and lead to drug resistance. The free energy decomposition analysis illustrates the importance of the loss of two important H-bonds in the L1152R and S1206Y mutants on drug resistance. The entropy analysis shows that the entropy term plays a critical role in the substantial change of the conformational entropies of G1202R and L1152R. Our results reveal the mechanisms of drug resistance and provide vital clues for the development of new inhibitors to combat drug resistance.