Spatial models are increasingly employed to help understand the distribution of organisms and establish conservation priorities. Classic patch-orientated models may have limited power to accurately predict the organisms' distributions. Pond breeding amphibians are appropriate study organisms because of their complex life cycle, low dispersal and sensitivity to environmental conditions. Here connectivity metrics and niche modelling were used to predict the occurrence of the northern crested newt in a rural landscape from central Romania. Pond-related variables, such as macrophyte cover and the presence of predatory fish, were the most important predictors of newt occurrence, followed by one landscape-related variable (urbanization) and a connectivity metric (nearest neighbouring occupied pond). Most of the landscape and connectivity variables were not adequate predictors, presumably because most of the terrestrial habitats in this traditionally used rural landscape are ecologically optimal for amphibians. Conservation measures for the northern crested newt should promote the preservation of traditional extensive agricultural practices and discourage stocking of ponds with predatory fish.