Abstract: For the studies of the total-load movable-bed model tests, the design of the model sand has become a key to the whole model testing. As the prototype sediment particle size is much small, ignoring the temperature differences between the model sand and the prototype sand, it is difficult to get the suitable natural model sand if the particle size of the natural sand calculated by the model scales is too small. So by controlling the flow temperature for the model experiments, under the conditions of different temperatures and different water flow viscosities, the reasonable sediment particle size can be got, so as to select optimized model sand in conditions satisfying suspension similarity. The sediment settling velocities and sediment settling scales have been simulated under the conditions of original and variable temperatures. And the model testing analysis results show that the choice of the model sand particle size in variable temperature conditions is available, and that the natural coarse sand used for simulation of the prototype finer sand not only meets the similarity requirements of the settling velocities, but the suspension similarity as well. The results of model testing validation indicate that the model sand selection has met the requirements of design and that the model design is reasonable.