The climate simulation of a fine-resolution (about 40 km) atmosphere-only climate model, the Australian Community Climate and Earth-System Simulator (ACCESS) with specified observed AMIP-style SSTs is evaluated to determine its utility for conducting experiments on the relationship between climate and tropical cyclone formation. Annual simulated numbers of tropical cyclones are good, with the model producing almost as many tropical cyclones as observed. There are regional variations in the quality of the model simulation, with fewer than observed cyclones simulated over the Atlantic basin, for instance. The model is then run with various representations of SST patterns and topographies, ranging from purely aquaplanet simulations with tropical SST values everywhere, through to aquaplanet simulations with various specified meridional SST gradients, to Earth topography simulations with zonally-averaged SSTs. The motivation for this work is to try to quantify the relationship between climate and tropical cyclone formation using a graded series of experiments that bridge the gap between the current terrestrial climate, where about 80 tropical cyclones are simulated per year, to a “tropical cyclone world” with no land and tropical SSTs everywhere, where many more tropical cyclones are generated. The constant tropical SST results show that as SST is increased, the number of tropical cyclones decreases. This is likely related to the decrease in the lapse rate with increasing SST, associated with an increase in precipitation for higher SST values causing increased latent heat release in the upper troposphere. The results of other SST experiments are also discussed and functional relationships between climate variables and tropical cyclone formation rates are proposed.