This article describes how natural geothermal soil temperature gradients in Iceland have been used to study terrestrial ecosystem responses to soil warming. The experimental approach was evaluated at three study sites in southern Iceland; one grassland site that has been warm for at least 50 years (GO), and another comparable grassland site (GN) and a Sitka spruce plantation (FN) site that have both been warmed since an earthquake took place in 2008. Within each site type, five ca. 50 m long transects, with six permanent study plots each, were established across the soil warming gradients, spanning from unwarmed control conditions to gradually warmer soils. It was attempted to select the plots so the annual warming levels would be ca. +1, +3, +5, +10 and +20 °C within each transect. Results of continuous measurements of soil temperature (Ts) from 2013-2015 revealed that the soil warming was relatively constant and followed the seasonal Ts cycle of the unwarmed control plots. Volumetric water content in the top 5 cm of soil was repeatedly surveyed during 2013-2016. The grassland soils were wetter than the FN soils, but they had sometimes some significant warming-induced drying in the surface layer of the warmest plots, in contrast to FN. Soil chemistry did not show any indications that geothermal water had reached the root zone, but soil pH did increase somewhat with warming, which was probably linked to vegetation changes. As expected, the potential decomposition rate of organic matter increased significantly with warming. It was concluded that the natural geothermal gradients at the ForHot sites in Iceland offered realistic conditions for studying terrestrial ecosystem responses to warming with minimal artefacts.