The Kam Tin Main Drainage Channel (KTMDC) is an important river for the city drainage in Hong Kong. The roughness and its variations have an obvious effect on the flood control capacity and the flow capacity. So physical model tests are designed to study the KTMDC. Due to its complex channel structure, the tests are completed in two steps. In Step 1, the energy loss is measured along the main channel without inflows, with all inflows and outflows being sealed. In Step 2, all the inflow and outflow structures are measured, with the sealed inflows and outflows being opened on the basis of Step 1. In each step, two schemes are employed. One of the key issues is the choice of suitable materials to make the model’s roughness similar to that of the prototype. According to the gravity similarity criterion, the 1:25 scale model is built, with the main channel made of Perspex. The facing slopes of the grasscrete and the stone masonry need to be roughened. A kind of the nylon net is selected to simulate the roughness of the stone masonry and the plastic lawn for the grasscrete facing slope. For the different structure reaches, the roughness coefficients are estimated based on the hydraulic theory. The rationality of the test results is verified in this study. The results of testing can provide a reliable basis for the renovation, the expansion, the optimization of this channel. The Kam Tin Main Drainage Channel (KTMDC) is an important river for the city drainage in Hong Kong. The roughness and its variations have an obvious effect on the flood control capacity and the flow capacity. So physical model tests are designed to study the KTMDC . Due to its complex channel structure, the tests are completed in two steps. In Step 1, the energy loss is measured along the main channel without inflows, with all inflows and outflows being sealed. In Step 2, all the inflow and outflow structures are measured, with the sealed inflows and outflows being opened on the basis of Step 1. In each step, two schemes are employed. One of the key issues is the choice of suitable materials to make the model’s normal similar to that of the prototype. According to the gravity similarity criterion, the 1:25 scale model is built, with the main channel made of Perspex. The facing slopes of the grasscrete and the stone masonry need to be roughened. A kind of the nylon net is selected to simulate the roughness of the stone masonry and the plastic lawn for the grasscrete facing slope. For the different structure reaches, the roughness coefficients are estimated based on the hydraulic theory. The rationality of the test results is verified in this study. can provide a reliable basis for the renovation, the expansion, the optimization of this channel.