High-precision　and　high-resolution　satellite　altimetry　data　from　CryoSat-2　are　widely　utilized　for　marine　gravity　inversion.　The　vertical　gravity　gradient　is　a　crucial　parameter　of　the　Earth＇s　gravity　field.　To　evaluate　the　performance　of　vertical　gravity　gradient　determined　from　CryoSat-2　altimeter　data,　the　pre-processed　along-track　sea　surface　heights　(SSHs)　are　obtained　through　error　correction.　The　study　area　focused　on　the　Arabian　Sea　and　its　surrounding　region,　where　the　along-track　geoid　was　derived　by　subtracting　the　mean　dynamic　topography　of　the　ocean　from　the　along-track　SSH　of　CryoSat-2.　The　residual　along-track　geoidal　gradients　were　obtained　by　adjusting　the　along-track　geoid　gradients　calculated　from　CryoSat-2　altimeter　data　using　the　remove-restore　method.　This　was　done　by　subtracting　the　geoid　gradients　calculated　by　the　gravity　field　model　XGM2019e_2159.　After　obtaining　the　residual　along-track　geoidal　gradients,　the　residual　gridded　deflections　of　the　vertical　(DOV)　are　calculated　using　the　least-squares　collocation　(LSC)　method.　The　residual　gridded　DOV　are　then　used　to　compute　the　residual　gridded　gravity　anomaly　gradients　in　the　study　area　using　the　finite-difference　method.　After　restoring　the　gravity　anomaly　gradients　computed　by　the　XGM2019e_2159　model,　a　high-resolution　gravity　anomaly　gradient　model　with　a　resolution　of　1′　×1′　is　obtained　for　the　Arabian　Sea　and　its　surrounding　area.　To　evaluate　the　accuracy　of　the　gravity　anomaly　gradient　model　derived　from　CryoSat-2,　it　was　compared　with　the　SIO　V32.1　gravity　anomaly　gradient　model　released　by　the　Scripps　Institution　of　Oceanography.　The　comparison　showed　that　the　root　mean　square　(RMS)　of　the　differences　between　the　two　models　is　7.69E,　demonstrating　the　high　accuracy　and　precision　of　the　vertical　gravity　gradient　determined　from　CryoSat-2　altimeter　data.　©　2023　The　Author(s).　Published　by　Oxford　University　Press　on　behalf　of　The　Royal　Astronomical　Society.