Friction is a complex and important property for almost all forming processes, which influences a large number of process variables. This is especially true in bulk metal forming, as friction has, for example, significant influence on the forming force. Furthermore, knowledge of friction conditions is important for resource-saving production in order to save material and energy through more efficiently designed production processes. To describe the interaction between tool and workpiece, friction models have been developed which mathematically model the frictional stresses for the simulation of different forming processes. In bulk metal forming, Tresca model is often used due to high occurring normal stresses. However, the model may not be sufficient when areas of low normal stresses and areas of high normal stresses occur simultaneously, since the model only considers flow stress of the contact partners and neglects the influence of normal stresses. For these cases, combined Coulomb-Tresca model can be used instead. The challenge for model calibration is that friction can hardly be measured directly and therefore is often measured indirectly by quantities that are sensitive towards friction, using laboratory experiments. Measured quantities are often geometric, such as the change in outer contour in the present case. In this work, conical tube-upsetting tests, matching process conditions of hot bulk metal forming, are used for friction model calibration. A methodology is presented that allows for determination of both model parameters of the combined Coulomb-Tresca model simultaneously, using an inverse approach. In a first step, a new specimen centering and data processing procedure are established, which allow for robust and reproducible tests. During testing, a line-laser is used to measure the shape of the specimen’s outer contour continuously over the process. In a second step, friction model parameters are determined by minimizing the difference between numerically calculated and experimentally measured outer contour of the specimen by means of mathematical optimization. The inversely determined parameter set is finally validated using a cup-backwards extrusion process.