Dual-coil frequency-domain electromagnetic induction (EMI) systems are commonly used as detectors of buried metallic objects, but they are also increasingly used for environmental purposes such as detection of contaminant plumes and archaeological prospection. Usually, data are analyzed directly by visualizing the in-phase and quadrature components, and also by applying one-dimensional inversion methods. Besides, there exist three-dimensional (3-D) forward and inverse modeling codes based on finite-difference techniques, but these methods are not routinely applied because their computation cost for real geophysical situations is still too high. The computation cost is significantly lower for two-dimensional (2-D) structures since this problem is not 3-D but 2.5-D. Few 2.5-D methods have been published in the last years, based on finite-element techniques, but for the case of electric dipole sources. In this paper, the authors present a 2.5-D forward-modeling algorithm, based on Rayleigh-Fourier expansions, for calculating the response of 2-D multilayered earth with irregular boundaries to the magnetic-dipole sources. Using this code, the authors numerically simulated the dual-coil frequency-domain EMI response of a soil model that could be found in environmental research. They considered a buried nonmetallic object, conductive with respect to the host media, and calculated its response for different orientations of the transmitter and receiver coils. The best resolution for detecting and characterizing this object corresponded to the configuration in which the axes of the transmitter and receiver dipoles were parallel to the ground surface and perpendicular to the symmetry axis of the buried objects, and the axis of the instrument was parallel to that symmetry axis. Finally, the authors interpreted the field data from a profile exhibiting resistive anomalies, corresponding to underground contamination, by using their forward code and a trial-and-error procedure. This profile had been previously characterized through the inversion of dipole--dipole electrical data. They considered that result to select their starting multilayered model. They obtained a good correlation between the EMI data and the synthetic response of the final multilayered model. Besides, this model is consistent with the image of the electrical inversion. During the modeling process, the method showed to be practical and versatile and to have a good convergence. Index Terms--Discrete Fourier transforms, dual coil, electromagnetic induction (EMI), forward modeling, frequency domain, Rayleigh scattering, two-dimensional structures.