In most marine reclamation projects, sand fill is placed directly on soft marine seabed soils. The sand particles can easily penetrate into the soft marine soils, and the soft soil can also move into the pore spaces inside the sand at the initial contact interface between the sand and the soft marine soil. In this case, the permeability and the volume of the sand above the initial surface are reduced. To avoid this problem, a geotextile separator is often placed on the surface of the soft marine soils before placing the sand. In this study, a two-dimensional physical model is utilized to study the geotextile separator effects. The initial conditions of a clayey soil, sand fill, and surcharge loading were kept the same in the physical model test with the only difference being that a geotextile separator was either placed on the clay surface or omitted. The settlements of the initial interface were recorded and compared for the two cases without or with the geotextile separator. The particle size distribution of the soils taken across the interface zone for different time durations was then measured, analyzed, and compared. Based on an analysis of the results, the sand percolation depth was 40 mm and fine particle suffusion was apparent when the sand was placed directly on the marine slurry surface without a geotextile separator. However, when a geotextile separator was used sand percolation was avoided, and the fine particle suffusion was effectively diminished. A relative fine particle fraction is defined to illustrate the migration of fine particles from the clay to the sand soils. The fine particle percentages of the Hong Kong Marine Deposits–sand mixtures were calculated for the cases with and without a geotextile separator using an empirical formula and micromechanical modeling to obtain a better understanding of the effects of geotextile separators in practice.