The effect of primary shape parameters, such as side wall angle, offset distance and splitter distance, on the internal flow in classical wall-attachment devices has been examined experimentally at a Reynolds number of 16×(10)^3. The results show how the reattachment distance, pressure distribution and velocity profile in the device are affected by the shape parameters. In particular, the influence of splitter distance was examined in detail. The reattachment distance decreases continuously from a value in the case of without a splitter to a minimum as the splitter distance decreases from the fully far downstream distance. In contrast, the maximum value of the reattachment distance becomes partially and slightly larger than that without a splitter. Thus, the effective range of splitter distance where the reattachment distance varies from the maximum equivalent almost to the value without a splitter to the mimimum is determined as the range where the ordinary wall-reattachment flow pattern is directly affected by a splitter. It is shown that the effective range of reattachment and splitter distances nondimensionalized by the correctly defined upper-limit values becomes similar for all devices reported here.