Background: ‘Hours per patient day’ (HPPD) is an internationally recognised resourcing metric used to measure direct nursing care hours. However, hospitals often underestimate indirect time (unavailability) and specify unrealistic targets for planned unavailability (‘headroom’). Aims: To investigate the disparities between planned unavailability (‘headroom’) and actual staff unavailability. Methods: Data were collected from the e-rostering systems of 87 NHS trusts. This was compared with published data from 35 roster policies. Results: Many hospitals use headroom as a key performance indicator (KPI) and set targets for its components in their roster policies. This research highlights large variations in unavailability (15.8% to 33.6%) and lower variations in headroom (16–26%). Conclusion: Hospitals operationalise headroom around an idealised ‘target’ value. This may be detrimental. Compelling a unit with unavailability of between 28% and 30% to adopt an institution-wide headroom of 22% (for example) may, at best, increase spending on bank/agency staff, or, at worse, jeopardise patient safety.