Large scale magnetic fields are the fundamental key ingredient for both jet collimation and production. However, their presence is still a mystery: are they generated by a dynamo or simply advected? We start by quickly reviewing the physics of jet launching from quasi keplerian (thin) discs. It will be shown that the only available solutions of powerful jets require fields close to (but smaller than) equipartition. All these solutions rely on a crude phenomenological local description of turbulence (alpha prescription). However, despite this caveat, they seem to be consistent with several observational constraints such as mass loss rates and kinematics of jets from young stars. The interplay between turbulent diffusion and advection is then addressed by 2.5 D simulations done with the MHD code PLUTO. It is shown that advection can indeed occur above a threshold on the disc magnetization. Our results shed a new light on the long term evolution of accretion discs by putting more emphasis on the magnetic history of each particular object. This has strong implications on the understanding of hysteresis cycles seeing in X-ray binaries. © 2010 American Institute of Physics.