The understanding of the mechanisms responsible for particle transport is of the utmost importance for magnetized fusion plasmas. A peaked density profile is attractive to improve the fusion rate, which is proportional to the square of the density, and to self-generate a large fraction of non-inductive current required for continuous operation. Experiments in various tokamak devices (AUG, DIII-D, JET, TCV, TEXT, TFTR) have indicated the existence of an anomalous inward particle pinch. Recently, such an anomalous pinch has been unambiguously identified in Tore Supra very long discharges, in absence of toroidal electric field and of central particle source, for more than 4 minutes [1]. This anomalous particle pinch is predicted by a quasilinear theory of particle transport [2], and confirmed by non-linear turbulence simulations [3] and general considerations based on the conservation of motion invariants [4]. Experimentally, the particle pinch is found to be sensitive to the magnetic field gradient in many cases [5,6,7,8], to the temperature profile [5,9] and also to the collisionality that changes the nature of the microturbulence [10,11,12]. The consistency of some of the observed dependences with the theoretical predictions gives us a clearer understanding of the particle pinch in tokamaks, allowing us to predict more accurately the density profile in ITER.