The influence of two nozzle geometries and three process parameters (arc current, arc length and plasma sheath gas flow rate) on the energy distribution for an argon transferred arc is investigated. Measurements are reported for a straight bore cylindrical and for a convergent nozzle, with arc currents of 100 A and 200 A and electrode gaps of 10 mm and 20 mm. These correspond to typical operating parameters generally used in plasma transferred arc cutting and welding operations. The experimental set up consisted of three principal components: the cathode-torch assembly, the external, water-cooled anode, and the reactor chamber. For each set of measurements the power delivered to each system component was measured through calorimetric means, as function of the arc's operating conditions. The results obtained from this study show that the shape of the cathode torch nozzle has an important influence on arc behaviour and on the energy distribution between the different system components. A convergent nozzle results in higher arc voltages, and consequently, in higher powers being generated in the discharge for the same applied arc current, when compared to the case of a straight bore nozzle. This effect is attributed to the fluidynamic constriction of the arc root attachment, and the consequential increase in the arc voltage and thus, in the Joule heating. The experimental data so obtained is compared with the predictions of a numerical model for the electric arc, based on the solution of the Navier-Stokes and Maxwell equations, using the commercial code FLUENT (c). The original code was enhanced with dedicated subroutines to account for the strong temperature dependence of the thermodynamic and transport properties under plasma conditions. The computational domain includes the heat conduction within the solid electrodes and the arc-electrode interactions, in order to be able to calculate the heat distribution in the overall system. The level of agreement achieved between the experimental data and the model predictions confirms the suitability of the proposed, "relatively simple" model as a tool to use for the design and optimization of transferred arc processes and related devices. This conclusion was further supported by spectroscopic measurements of the temperature profiles present in the arc column and image analysis of the intensity distribution within the arc, under the same operating conditions.