Today incremental bulk metal forming is applied in various industrial manufacturing processes. With this technology, high product quality and flexibility can be achieved. In this study, the rolling of spur gears and rims is investigated, using a new approach. With these processes, a ring- or cup-shaped work piece is rolled into a toothed mandrel, by which gears and rims in net-shape quality can be produced. For internal tooth systems flow forming is used, while for external tooth systems ring rolling is suggested. The FE simulation of flow forming processes for internally geared wheels is compared to experimental results showing a relatively close match of strain hardening effects. The simulation also shows the limitations of available finite element codes. Incremental forming processes are characterized by a relatively small forming zone moving through the work piece. Thereby similar forming steps reappear during the manufacturing. These specific properties of incremental processes are used to improve the efficiency of their numerical simulation. Since the simulation with conventional finite element methods is still very time-consuming, a new algorithm was developed and implemented into the FE-code LARSTRAN. These algorithms are based on the transformation and interpolation of the solution. The mechanical solution will be stored at each time step. If the similarity exists between the current and the preceding archived time steps, the transformation and interpolation will be performed. The results from transformation and interpolation are then checked by means of physical constraints. If the results pass the latter check, they will be archived as solution. Consequently, at this time step, no iterative computation will be accomplished by mechanical solver and as a result, the number of iterations will be reduced. Then the calculations by thermo-solver are accomplished conventionally in dependence on the archived mechanical results. With the help of these algorithms, the number of iterations will be reduced obviously and the results will be reasonably acceptable. Computational time can significantly be saved if the similarity of the forming steps is used. This is revealed by a simplified incremental process.