In this paper, we study the superconvergence of the error for the local discontinuous Galerkin (LDG) finite element method for one-dimensional linear parabolic equations when the alternating flux is used. We prove that if we apply piecewise k-th degree polynomials, the error between the LDG solution and the exact solution is (k + 2)-th order superconvergent at the Radau points with suitable initial discretization. Moreover, we also prove the LDG solution is (k + 2)-th order superconvergent for the error to a particular projection of the exact solution. Even though we only consider periodic boundary condition, this boundary condition is not essential, since we do not use Fourier analysis. Our analysis is valid for arbitrary regular meshes and for P k polynomials with arbitrary k ≥ 1. We perform numerical experiments to demonstrate that the superconvergence rates proved in this paper are sharp.Mathematics subject classification: 65M60, 65M15.The DG method was first introduced in 1973 by Reed and Hill [25], in the framework of neutron linear transport. Later, the method was applied by Johnson and Pitkäranta to a scalar linear hyperbolic equation and the L p -norm error estimate was proved [23]. Subsequently, Cockburn et al. developed Runge-Kutta discontinuous Galerkin (RKDG) methods for hyperbolic conservation laws in a series of papers [16][17][18][19]. In [20], Cockburn and Shu first introduced the LDG method to solve the convection-diffusion equation. Their idea was motivated by Bassi and Rebay [8], where the compressible Navier-Stokes equations were successfully solved.The superconvergence properties have been analyzed intensively. In [2, 5], Adjerid et al. studied the ordinary differential equations and proved the (k + 2)-th order superconvergence of the DG solutions at the downwind-biased Radau points. For hyperbolic equations, the superconvergence results have been investigated by several authors [6,7,9,10,12,24,26,27]. Especially, in [26], we obtained sharp superconvergence for linear hyperbolic equations by using the dual argument, and this gives us the motivation to the prove the sharp superconvergence for linear parabolic equations. For convection-diffusion problems, in [3,4], the authors used numerical experiments to demonstrate the superconvergence of LDG solution at the Radau points. In [11], the steady state solution was studied and the superconvergence of the numerical fluxes was proved. In [13], Cheng and Shu discussed the superconvergence property of the LDG scheme for heat equation by using piecewise linear approximations and uniform meshes. Subsequently, they proved the (k + 3 2 )-th order superconvergence when using piecewise k-th degree polynomials with arbitrary k on arbitrary regular meshes in [14]. However, the convergence rate obtained in [14] is not sharp. Numerical tests demonstrated that the error of the DG solution towards a particular projection of the exact solution is (k + 2)-th order accurate, even on highly nonuniform meshes. In [14], the framework to prove the superconvergence res...