We do not believe that the minimum energy theory is applicable to driven plasmas. We assume that relaxation occurs towards a state of minimum entropy production rate. As the equilibrium Ohm's law cannot be satisfied in peaked-current density states, the final state must be steadily turbulent. What is new about our theory is, fundamentally: (i) the pressure balance is considered as a relevant constraint upon the relaxed state and this is expressed as an integral relation; (ii) the energy dissipation by turbulence is assumed to be minimal and (iii) anisotropic resistivity is regarded as the main cause of poloidal current. The numerical calculations of magnetic field and current density reproduce typical observations in tokamaks.