Thermal plasma processes are now currently used in industry for many applications, each of them using adapted plasma torches or transferred arcs. However, many difficulties still exist in understanding what happens at the electrodes of direct current (dc) arcs. For the latter the material composition is continuously evolving and the arc root behaviour is a three-dimensional transient phenomenon. For RF discharges, the coupling between the plasma and the discharge is much better understood.To model such flows, knowledge of the transport properties is a prerequisite and many works are in progress in this field. The difficulty in modelling lies in the fact that the plasma core is laminar while its fringes are turbulent. Moreover, the turbulence is more of the engulfment type than represented by k-ε, R ij . . . models and the phenomena are even more complex with arcs due to the piston flow generated by arc root fluctuations. Mixing a cold gas with the plasma is the key of many processes. Here also three-dimensional codes are now currently used but it is still difficult to introduce the transient behaviour of the plasma and calculate, fast enough, the-out-of-equilibrium composition and transport coefficients of complex mixtures.Models have been backed by measuring techniques: emission or absorption spectroscopy, enthalpy probes of reduced sizes (d ∼ 3 mm) coupled with a mass spectrometer, laser scattering techniques, fast imaging, with or without laser illumination, laser anemometry, fast (50 ns) pyrometry for particles in flight. Such techniques give a much better idea of local and/or transient phenomena, but they also confirm the need for the development of new models.