Forest cover change is a major contributing factor to global environmental change. Whereas several studies have focused on the general land use and land cover dynamics, we focus on analysing forest cover change patterns in a protected landscape taking into consideration how other land categories are increasing at the expense of the forest. In this study, we analyse forest cover change patterns and associated proximate land use factors between 1987 and 2017 using Landsat images from the Tano-Offin Forest Reserve (TOFR) in Ghana. Using the Random Forest machine learning algorithm, we classified the images into forest, developed land, and agricultural land. The study finds that forest cover losses are 1.9 and 1.4 times the amount of forest cover gains in 1987-2002 and 2002-2017, respectively. We find that even though the forest cover is more likely to recover from the agricultural land, land developers mostly targeted the agricultural land. The focus of Ghana's Forest and Wildlife Policy and the underlying process of forest cover change in the TOFR suggest that a country's forest policy should focus on a combination of diverse and spatially explicit proximate factors that are likely to threaten the integrity of forests.