To investigate agreement between hospital staff on what constitutes a fall and should be recorded on an incident report, to identify factors that influence whether a scenario is classified as a fall, and to examine the effect of providing a definition of a fall on interrater agreement. DESIGN Two interrater agreement trials with pre-post-intervention design SETTING Seven hospitals PARTICIPANTS: Four hundred forty-six hospital staff (nursing 76 , physiotherapy 14 , occupational therapy 6 ) working on participating wards over a 24-hour period. MEASUREMENTS: Fourteen video scenarios were played before participants, who were asked whether they would classify each scenario as a fall and whether they would complete an incident report in that scenario. A definition of a fall was then provided and the video scenarios replayed. Two definitions were investigated; the World Health Organisation definition and a custom definition. RESULTS: Consensus was not evident in five of the 14 scenarios examined. Falls onto surfaces higher than the ground were less likely to be recorded on incident reports. Hospital and ward type influenced whether a scenario was classified as a fall. Overall agreement in classifying scenarios as a fall was only marginally greater after provision of the definition in both phases, although neither had an effect on whether staff would complete an incident report. CONCLUSION: Disagreement between hospital staff in what constitutes a fall may contribute to inconsistent research findings in this field and is not meaningfully improved through provision of a definition.