At present, the dominant technology for transducers in the field of Ultrasonic Non-Destructive Testing is piezoelectric. However, some industrially important applications, like the inspection of components operating at high temperature or while in motion, are difficult tasks for standard piezoelectric probes since mechanical contact is required. In these cases, contactless NDT techniques can be an attractive alternative. Among the available options, Electromagnetic Acoustic Transducers (EMATs) can generate and detect ultrasonic waves without the need for a physical contact between the probe and the test object, as their operation relies on electromagnetic, rather than mechanical coupling. Since EMATs do not require any coupling liquid, the experimental procedures for inspection set-up are simplified and a source of uncertainty is eliminated, yielding highly reproducible tests that make EMATs suitable to be used as calibration probes for other ultrasonic tests. A further advantage of EMATs is the possibility of exciting several wave-modes by appropriate design of the transducer. Unfortunately, EMATs are also characterized by a relatively low signal-to-noise ratio and by a complex operation relying on different transduction mechanisms that make their performance dependent on the material properties of the testpiece. The present work aims to develop a numerical model including the main transduction mechanisms, the Lorentz force and magnetostriction, that can be employed as a prediction tool to improve the understanding of EMAT operation. Following an overview on the historical development of EMATs and their models, the theory describing EMAT operation is presented. The governing equations are implemented into a commercial Finite Element package. The multi physics model includes the simulation of the static and dynamic magnetic fields coupled to the elastic fields through custom constitutive equations to include magnetostriction effects. The model is used to quantitatively predict the performance of a magnetostrictive EMAT configuration for guided waves without employing arbitrary parameters. The results are compared to experimental data providing a validation of the model and insight on the transduction process. The validated model, together with experimental tests, is exploited to investigate the performance of different EMAT designs for Shear Horizontal waves in plates. The sensitivities of each configuration are compared and the effect of key design parameters is analyzed. Finally, the model is used in the evaluation of the performance of bulk wave EMATs on a wide range of steel grades. Experimental data interpreted via numerical simulations are employed to investigate the relative weight of the transduction mechanisms, with implications on the applicability of EMATs on the range of steels usually encountered in inspections.