<p>The Greenland Ice Sheet is a major contributor to global sea-level rise, with accelerating mass loss due to climate change. Accurate estimation of Greenland Ice Sheet mass variations is critical for understanding ice sheet dynamics and predicting future sea-level changes. However, spherical harmonic coefficient solutions from GRACE/GRACE-FO suffer from high-frequency noise and signal leakage, particularly at Greenland Ice Sheet-ocean boundaries, limiting their reliability in regional mass balance studies. In this study, we propose an Improved Slepian Method to address these limitations. The Improved Slepian Method refines the inversion strategy by introducing satellite-altitude pseudo-observations to separate Greenland Ice Sheet and surrounding ocean/island signals, recovering leaked signals. It also uses satellite altimetry data as a regularization matrix to constrain spatial patterns that are critical to reduce signal leakage. Additionally, it weights the data with the error covariance matrix to suppress high-frequency noise, enabling the use of higher-degree (e.g., degree 96) time-varying gravity field models. Validated against Input–Output Method, the Improved Slepian Method improves accuracy by 25–58% across the Greenland Ice Sheet and its drainage systems, outperforming the Classical Slepian Method. The Improved Slepian Method also demonstrates comparable performance to Mascon solutions in estimating long-term trends, while the Classical Slepian Method underestimates the trend by approximately 80 Gt/year due to signal leakage. This study addresses the issues of high-frequency noise and signal leakage in spherical harmonic coefficient solutions, providing a robust framework for high-precision Greenland Ice Sheet mass balance research and other geophysical applications.</p>