Sea ice numerical model is an effective way to study the dynamical and thermal state parameters of sea ice and the connections between them. The current assessment of the results of numerical ice thickness simulation is much less than the sea ice extent/area and concentration, and the study of the influence of ice velocity and sea ice deformation on ice thickness distribution is still lacking. We simulated the Arctic sea ice variability from 1980 to 2018 using the Los Alamos sea ice model (CICE), and validated and comparison the CICE simulation results using remote sensing and assimilated ice thickness data. We further analyzed the effects of simulated ice velocity and sea ice deformation on ice thickness, and calculated the contributions of ice velocity divergence and shear bias to ice thickness bias. The results show that the interannual variability of the mean ice thickness and ice speed in the Arctic north of 70°N is reasonable, but the multi-year trends of the simulated mean ice thickness and ice speed are smaller than the variability of the assimilated data; the differences in the spatial distribution of the simulated and observed ice thickness are closely related to the deviations of the ice speed and deformation rate, mainly in the positive deviation in the Beaufort Sea and the negative deviation in the Arctic central zone to Fram Strait. The contribution of divergence and shear deviation to ice thickness deviation in the pan-Arctic region fluctuates between 13% and 16% before March, and jumps from 16% to 27% in March-April. The divergence bias dominates the positive bias of ice thickness in the Beaufort Sea region in November and December, while the shear bias dominates the negative ice thickness bias in the winter in the ocean north of the Canadian archipelago and in the region of the Arctic transpolar drift.