Nested iterative solutions using full and approximate forms of the multilevel fast multipole algorithm (MLFMA) are presented for efficient analysis of electromagnetic problems. The developed mechanism is based on preconditioning an iterative solution via another iterative solution, and this way, nesting multiple solutions as layers. The accuracy is systematically reduced from top to bottom by using the on-the-fly characteristics of MLFMA, as well as the iterative residual errors. As a demonstration, a three-layer strategy is presented, considering its parametrization for accelerating iterative solutions of perfectly conducting objects. We show that the strategy significantly reduces the solution time, especially for ill-conditioned matrix equations that are derived from the electric-field integral equation.