In Namibia, leopards (Panthera pardus) are widely distributed, used commercially as trophy animals and are often persecuted for perceived or real predation on livestock and valuable game species outside protected areas. Therefore, leopard populations living in protected areas might be important source populations and for maintaining connectivity. Little data on their population sizes and densities are available from the northern part of the country, particularly from protected areas. Here, we estimated leopard densities using a spatial capture–recapture approach in northern Namibia: (i) the Khaudum National Park (KNP) in north-east Namibia with an annual average rainfall of 450 mm and (ii) the Lower Hoanib River (LHR) in north-west Namibia with an annual average rainfall of 25 mm. With an effort of 2430 and 2074 camera trap nights in the KNP and LHR, respectively, 11 adult female and six adult male leopards were identified in the KNP, whilst only one adult female leopard was detected once in the LHR. For the KNP, a maximum likelihood approach (using the package SECR) revealed a density estimate of 2.74 leopards/100 km2, whereas a Bayesian approach (using the package SPACECAP) revealed a density estimate of 1.83 leopards/100 km2. For the LHR, no density estimate could be determined and it is suggested that the leopard density in such an arid environment is low. These are the first leopard density estimates based on camera trap surveys provided for these protected areas and thus of importance for further monitoring programs to understand leopard population dynamics. We discuss our findings with current habitat changes and conservation measures in both study areas. Keywords Leopard, Panthera pardus, camera trap, density estimate, spatial capture-recapture, Namibia.