The invention discloses a non-uniform leakage correction method suitable for bar-by-bar calculation. The method comprises the following steps of 1, performing statistics on lattice cell information, firstly performing high-order transport calculation on non-uniform components, and then performing calculation to obtain uniform cross sections and average neutron flux density of all lattice cells, and net neutron current density and neutron flux density of the surface of each lattice cell in the components; 2, generating a discontinuous factor for a non-uniform leakage correction solving method according to an equivalent uniformization principle, and solving a non-uniform leakage correction model by adopting a longitudinal coordinate discretization method; and 3, correcting neutron flux density distribution by utilizing a critical energy spectrum generated by non-uniform leakage correction, performing re-merging to obtain the uniform cross sections, generating corresponding diffusion coefficients, and multiplying the uniform cross sections and the diffusion coefficients by flux ratios of uniform components and the non-uniform components of the lattice cells. According to the method, an energy spectrum correction factor and a leakage coefficient of each lattice cell can be obtained by utilizing the non-uniform leakage correction model.