Predicting the spatial variation of soil thickness is essential to improve our understanding different hydrological and geomorphological processes, such as landslides, surface runoff generation, and, consequently, erosional processes. The variable “soil thickness” is also crucial for engineering with a focus on excavations and the construction of underground structures. Despite its importance, estimating the spatial distribution of soil thickness in the field is not an easy task, especially in tropical areas with thick soil mantles and highly undulating topography. This study focuses on estimating the spatial distribution of soil thickness based on empirical models that use topographic parameters, such as elevation, hillslope angle, curvature, topographic position index, and topographic wetness index. The four models used were tested in the Papagaio river basin, in the city of Rio de Janeiro. Their parameterization and validation were implemented using data from 137 penetration tests carried out in the field with the usage of a light dynamic penetrometer (DPL), which indicates that the soil thickness may reach up to 15 m. The results provided information for a discussion concerning the strengths and limitations of each model. The best results were achieved by the model that considered the largest number of parameters to estimate soil thickness. The empirical models used here do not directly consider the erosional and depositional processes that occur in the landscape. Therefore, future studies should include more complex predictive models that are more adequate to the reality and complexity of this type of landscapes.