A practical and fast method for computing power systems' maximum loading point (MLP) using the load flow with step size optimization (LFSSO) technique is presented in this paper. The method is based on successive calculations of operating points in the infeasible region of the parameter space, from a large loading (larger than the actual maximum) down to the actual maximum loading, by using a load shedding technique based on the geometric characteristics of the maximum loading hyperplane. Also, a practical procedure is added to the process by including the system's operator information about the maximum admissible error allowed in the value of MLP. The model was also improved by considering the possibility of calculating the MLP for the whole system or for specific predefined areas. The method was applied to IEEE test systems and actual distribution and transmission systems, as the Brazilian system with more than 4,500 buses. The MLP is obtained after a few iterations, showing that the proposed method is very efficient. © 2013 IEEE.