A computer code, that couples numerical modelling and optimization routines, is described and used to automatically optimize the performance of a multi-step calibration stage of a plastic profile extrusion line. The objective of the optimization process is to minimize two conflicting criteria: final plastic profile average temperature and corresponding distribution standard deviation, both computed at the end of the calibration/cooling stage. The optimization routine uses Multi-Objective Evolutionary Algorithms (MOEA). Having in mind the well-known difficulties in characterizing the interface heat transfer coefficient, the system optimization was performed using different constant values for this parameter.