Batteries are used in a wide variety of applications as an energy store to bridge gaps in the primary source of supplied power for a given period of time. In some cases this bridging time, the battery's "autonomy", is fixed by local legislation but it is also often set by historically common practices. However, even if common practice dictates a long autonomy time, we are entering a new era of "cost and benefit realism" underpinned by environmentally friendly policies and we should challenge these historical practices at every opportunity if it can lead to resource and cost savings. In some cases the application engineer has no choice in the design autonomy; either follow a piece of local legislation (e.g. 4 h autonomy for a "life safety" application), or actually work out what is needed! An example of the latter would be for a remote site, off-grid, using integrated wind/solar power (without emergency generator back-up) where you may have to design-in several days' battery autonomy. This short paper proposes that a battery's autonomy should be related to the time expected for the system to be without the primary power source, balanced by the capital costs and commercial risk of power failure. To discuss this we shall consider the factors in selecting the autonomy time and other related aspects for high voltage battery systems used in facility-wide uninterruptible power supply (UPS) systems. (C) 2004 Published by Elsevier B.V.