Hybrid and electric vehicles are increasingly gaining importance in the automotive development. Like all industrially manufactured products, electric drives are subject to deviations within specified tolerance boundaries. Despite these deviations, the accuracy of torque and power must satisfy high standards. The effect of these deviations on flux and torque of the electric drive is analyzed in this thesis. The investigations focus on inverter-driven permanent magnet synchronous machines (PMSM). The main objective is to compensate for torque error resulting from such deviations by inverter software. A sensitivity analysis is conducted to investigate the impact of specific production related deviations on the controlled system. Therewith, two main influence factors are identified: The magnet's remanence and the air gap thickness. Based on this, multiple compensation algorithms for increasing torque accuracy are developed. As a precondition for successful compensation, the deviation of each individual machine has to be detected. An identification procedure is introduced that quantifies key parameters of the individual machine's characteristic by simple measurements. For experimental validation, the concepts are tested on a dyno setup with dedicated machines. To determine the magnet's influence, limit sample machines with increased or decreased remanence are used. Another limit sample machine features an increased air gap thickness. Driving these machines without compensation, a significant torque error is feasible. With the presented identification methods, the deviation of the electric machine in each individual vehicle can be quantified. By applying the developed compensation algorithms, its torque can be controlled very precisely despite considerably large deviations. With this, the torque error can be halved or lowered even more.