Projecting future sea-level rise is critical for mitigation and adaption planning in many highly-populated coastal areas. The response of sea level to anthropogenic climate change is not spatially uniform, with different regions experiencing higher or lower sea-level change than the global average. Ocean dynamics is one of the important factors to cause non-uniform regional sea level distribution, which is referred to as the dynamic sea level. In this study, we analyse dynamic sea level simulations from the ensemble of Coupled Model Intercomparison Project Phase 6 (CMIP6) models, focusing on the following three aspects. First, we evaluate representations of dynamic sea level in CMIP6 models by comparing its mean and variability on different timescales (seasonal, interannual, and interdecadal) against available observations. Second, we derive dynamic sea level projections for the coming centuries and identify similarities and differences across emission scenarios and underlying mechanisms. Third, we examine how model mean biases contribute to the uncertainties of regional sea level projections. The purpose of this study is to produce updated dynamic sea level projections based on CMIP6 models with better understanding of the uncertainties, which will be further incorporated into developing total sea level projections.