Multispectral scanners (MSS) onboard Landsat satellites 1-5 provide the earliest record of Earth observation data. Currently available Landsat MSS data products are processed by the U.S. Geological Survey (USGS) using the Level 1 Product Generation System (LPGS) to either L1T or L1G level of correction. MSS images processed to L1T have higher geometric accuracy since the images are precisely geo-referenced and orthorectified. The L1G MSS image product is available in cases when the elevation data and ground control information necessary to perform L1T correction are not available. The use of L1G MSS products for multitemporal change analysis necessitates accurate geometric correction of these images. This letter presents an automated method for accurate geo-referencing and orthorectification of L1G MSS images. The method is used to geometrically correct L1G MSS images acquired over South Dakota and the Himalayas, two areas representing vastly different landscapes. The performance of the method is evaluated quantitatively, and an average geometric accuracy of 0.47 pixel at MSS resolution is achieved. While the applicability of this method is demonstrated here using selected L1G MSS images, it has potential for accurate geometric correction of the entire archive of Landsat L1G MSS images.