Dimensional accuracy is a highly important quality criterion for injection molded parts and due to the decreasing tolerance criteria an ongoing issue for mold makers, part designers and manufacturers. Variations in the molding process regarding temperature, pressure and thus the specific volume are leading to inner stress distributions within the part, which ultimately cause warpage. To reduce warpage, a homogenization of the local melt properties, such as the temperature, cooling rate or specific volume is necessary. To be able to manipulate these melt properties locally, a mold with a highly segmented temperature control, consisting of 18 heating ceramics and carbon dioxide expansion chambers was designed. This work seeks to reduce warpage using a newly developed control loop that adjusts target values for the upcoming molding cycle while considering the control error of the previous cycle. The self‐optimizing control loop‐ aiming to homogenize the local mold temperatures over multiple cycles‐ was evaluated in experimental trials. Test specimens were produced, and measured for warpage. Compared to the reference (no control), the warpage of a plate‐shaped geometry was reduced by 10 % with the mold temperature homogenization approach.