This paper presents an experimental investigation on the impact of environmental variables on soil cracking in natural conditions. The test was performed on a large soil specimen of initial size 3 × 3 × 0.5 m exposed to real atmospheric conditions during one year, to include different seasonal weather conditions. The specimen was instrumented to monitor and record the main variables within the soil (temperature, volumetric water content, suction) and others at the soil-air interface (wind speed and direction, temperature, relative humidity, solar radiation, rain intensity). The experiment extends previous experience from laboratory desiccating tests in two main aspects: the size of the specimen and the exposition to a natural environment. A large size is desirable to reduce the effect of the mechanical boundary conditions. Conducting the test in a natural setting allows including variables that cannot be properly studied in laboratory conditions such as solar radiation or wind velocity, which this research has shown to constitute two key factors controlling water evaporation and eventually soil cracking. The soil-air interface constitutes a narrow zone with high gradients of most of the variables involved controlling the fluxes of water and energy. The experiment has shown that variation of those gradients has considerable implications in soil cracking.