A method is developed to compute the optimal multi-isotope separation in a gas centrifuge. The method relies on three models : Onsager’s pancake equation, diffusion equations written for each isotope, and an optimization routine. Onsager’s equation, well studied in the past for UF6, is adapted to multi-isotope gas mixtures, focusing on the three drives generating the countercurrent flow in practical centrifuges : feed drive, scoop drive, and linear wall thermal drive. Diffusion equations are written for each isotope in the initial form of partial differential equations (PDE) and reduced to ordinary differential equations (ODE) by the radial averaging method. These ODEs, linked with the solution of Onsager’s equation through two parameters (flow profile efficiency and scaled countercurrent flow), are solved by an iteration method. The optimization routine is based on a choice of a strategy and a method, using the solutions of the two preceeding models, to determine the optimal countercurrent driving parameters. Two examples of application, having an industrial interest respectively in the reprocessing of nuclear waste and in the production of stable isotopes, are presented : The reenrichment of spent reactor uranium and the separation of chromium isotopes.