We present modeling results of basic electron cyclotron heating (ECH) scenarios in several tokamaks and ITER performed with the most recent version of the three-dimensional (3-D) full-wave STELEC code (stellarator ECH, including tokamaks as a special case). This code includes all basic wave physics such as interference, diffraction, wave tunneling, mode conversion to electron Bernstein waves at the upper hybrid resonance (UHR), and appropriate boundary conditions. The code solves the wave equations in real 3-D magnetic geometry and thanks to the use of massive parallel teraflop computers, it is the first to provide full-wave solutions of the problem in toroidal plasmas. Several important new results are thus obtained that cannot be predicted with codes based on ray-tracing techniques, such as the influence of diffraction effects and the importance of the UHR for both X- and O-mode antenna excitation at fundamental harmonic. This last result also shows that the so-called "O and X" modes are coupled solutions. The coupling of these modes, partly supported by experiments in the DIII-D tokamak showing similar heating efficiencies for both radiated modes, leads to different power deposition profiles and spatial distribution, compared to results from ray-tracing codes. Coupling between the O-mode and the X-mode (launched at the low-field side) reveals the importance of electron Bernstein waves in ECH calculations for high-density ITER plasmas. These results not only could influence the predictions for neoclassical tearing mode suppression for ITER using electron cyclotron current drive but could also lead to important simplifications in ECH hardware (converters, polarization, etc.) and to a reduced cost of the ECH system on ITER. The code also allowed investigation of the urgent issue of the O-X-B ECH scenario for overdense tokamak/stellarator plasmas.