The evaporation rates of liquid hydrogen and oxygen spilled onto the ground surface were measured in laboratory tests. To simulate the ground, concrete, dry sand and wet sand layers were used in a vacuum-insulated cylindrical glass vessel. Based on the temperature variations within the layer and detailed observation through the side of the vessel, the heat transfer modes controlling the evaporation phenomenon were elucidated. When a wet sand layer was used, the liquid oxygen or hydrogen did not soak into the layer, because the frozen layer of water between the liquid and the sand layer acted as a barrier. When a concrete layer was used, the liquid vaporized above the layer. In these cases, the evaporation rates were inversely proportional to the square root of the time, except in the early stage just after the start of vaporization. This relationship could be predicted by a simple calculation of heat conduction within the layer. On the other hand, when a dry sand layer was used, liquid oxygen was observed to vaporize while constantly soaking into and rolling up the upper section of the layer. In this case, the evaporation rate was determined simply by the velocity of liquid penetration downward through the sand layer. When liquid hyrodgen was used, the liquid did not soak into the dry sand layer, and the evaporation mechanism seemed to be the same as that for the wet sand layer, because the air contained within interparticle cavities in the dry sand layer solidifies at the boiling point of liquid hydrogen.