Tracking is a very important issue for control and has received considerable treatment in the input affine framework. Unfortunately, many real systems exhibit a behavior which is clearly non input affine. To cope with this, an integral extension of the model can be used to produce an extended input affine system, hiding the input nonlinearity in the drift term of the new model. This approach, however, has several disadvantages. In this paper we consider a different approach, motivated by physical considerations and practical experiences, which is based on the transformation or approximation of the original system by a cascade of a nonlinear non input affine map and a following input affine nonlinear system. As the paper shows, well known results of tracking control can be easily extended to this class yielding the solution of the original problem. Simulation results are presented to confirm the validity of the proposed techniques. © 2005 AACC.