Controversy surrounds the averaging of visual acuity scores. We examine this debate and provide a series of guidelines and worked examples to enable investigators to select an average (e.g. arithmetic mean, geometric mean, median) appropriate for their data.