In the framework of the IAG African Geoid Project, an attempt towards a precise geoid model for Africa is presented in this investigation. The available gravity data set suffers from significantly large data gaps. These data gaps are filled using the EIGEN-6C4 model on a 15′× 15′ grid prior to the gravity reduction scheme. The window remove-restore technique (Abd-Elmotaal and Kühtreiber, Phys Chem Earth Pt A 24(1):53–59, 1999; J Geod 77(1–2):77–85, 2003) has been used to generate reduced anomalies having a minimum variance to minimize the interpolation errors, especially at the large data gaps. The EIGEN-6C4 global model, complete to degree and order 2190, has served as the reference model. The reduced anomalies are gridded on a 5′× 5′ grid employing an un-equal weight least-squares prediction technique. The reduced gravity anomalies are then used to compute their contribution to the geoid undulation employing Stokes’ integral with Meissl (Preparation for the numerical evaluation of second order Molodensky-type formulas. Ohio State University, Department of Geodetic Science and Surveying, Rep 163, 1971) modified kernel for better combination of the different wavelengths of the earth’s gravity field. Finally the restore step within the window remove-restore technique took place generating the full gravimetric geoid. In the last step, the computed geoid is fitted to the DIR_R5 GOCE satellite-only model by applying an offset and two tilt parameters. The DIR_R5 model is used because it turned out that it represents the best available global geopotential model approximating the African gravity field. A comparison between the geoid computed within the current investigation and the existing former geoid model AGP2003 (Merry et al., A window on the future of geodesy. International Association of Geodesy Symposia, vol 128, pp 374–379, 2005) for Africa has been carried out.