Early and accurate memory size estimation of Electronic Control Units (ECUs), which are dedicated embedded systems providing major software functionalities, is essential for automotive industry. Usually, experts using their expert judgments perform the estimations. Software functional size measures, which attempt at measuring the size by capturing the amount of functionality, have become popular in software industry particularly for effort estimation and project management purposes due to their availability early in the project lifecycle. This study investigates the potential usage of COSMIC Function Points as a predictor of memory size in comparison to expert opinion methods through conducting a case study in Renault.