Modular Battery Energy Storage systems (MBESSs) enable the use of lower-rated voltage converters and battery modules, and simpler battery management systems. They also improve the system's reliability and allow flexible power sharing amongst different modules. This paper proposes a power-sharing algorithm which maximizes the energy conversion efficiency of this battery energy storage system, considering state of charge (SoC) balancing and battery lifespan. Real time optimum power sharing is undertaken based on a simple look-up table, whose data was generated via off-line Genetic Algorithm optimization considering the converter's efficiency map. To demonstrate the proposed algorithms effectiveness, a six-module prototype system was constructed, each comprising of a halfbridge converter and a 10 Ah, 12.8 V, LiFePo4 battery. System testing occurred at different battery power levels in both charging and discharging modes, using the proposed efficiency optimized power-sharing and the conventional SoC-based powersharing methods. The results obtained show that the proposed power-sharing control significantly improves the light load efficiency compared to the conventional and equal power sharing methods. At high loads, the proposed method gives a higher efficiency than the SoC-based method, and an equivalent efficiency to the equal power-sharing method.