Modular UPS are usually presented as having an extremely high level of availability. In fact, the best modular UPS systems are designed to increase the global system's availability with minimum downtime due to their hot-swappable parts and cost-effective redundancy, the complete independence of each module and no fault propagation if one of the modules fails. However, the biggest concern is the internal reliability of UPS systems owing to the use of many modules. This paper explains and quantifies this issue and describes how to handle it, proposing design rules to significantly increase the modules' MTBF and giving a method for calculating and verifying it.