Abstract Space-based remote sensing of the region of specular reflection of sunlight from the ocean surface, termed sunglint, is useful for estimating the roughness characteristics of the reflecting ocean surface. Techniques have been developed to estimate near-surface wind speeds over the oceans based on a study of sunglint signatures in satellite images. The geostationary VHRR visible band imaging provides a veritable storehouse of sunglint signatures. In this paper, an algorithm for the precise location of the principal point of the specular solar reflection in INSAT-VHRR visible band images is presented and discussed. The principal point of specular reflection of the glitter pattern is estimated by the considering Sun-satellite-Earth geometry and the laws of reflection. This is achieved by knowing the position of the Sun and position of the satellite in the appropriate spherical coordinate system. The nonlinear equations thus derived are solved using the Newton-Raphson method. Using the algorithm, the location and movement of the principal point of specular reflection as a function of time of the day and day of the year is studied. An example of observed sunglint signatures in the INSAT-1B VHRR image of 7 August 1987 is analyzed and wind-speed estimation is presented. The algorithm presented in this paper is general and is applicable for the precise location and study of sunglint signatures in any other geostationary VHRR imaging sensors. The method presented also provides the basis for extension of this study to all other types of orbiting satellites.