BACKGROUND: Accidental falls are very common in older hospital patients -- accounting for 32% of reported adult patient safety incidents in UK National Health Service (NHS) hospitals and occurring with similar frequency in settings internationally. In countries where the population is ageing, and care is provided in inpatient settings, falls prevention is therefore a significant and growing risk-management issue. Falls may lead to a variety of harms and costs, are cited in formal complaints and can lead to claims of clinical negligence. The NHS Litigation Authority (NHSLA) negligence claims database provides a novel opportunity to systematically analyse such (falls-related) claims made against NHS organisations in England and to learn lessons for risk-management systems and claims recording. OBJECTIVES: To describe the circumstances and injuries most frequently cited in falls-related claims; to investigate any association between the financial impact (total cost), and the circumstances of or injuries resulting from falls in "closed" claims; to draw lessons for falls risk management and for future data capture on falls incidents and resulting claims analysis; to identify priorities for future research. METHODS: A keyword search was run on the NHSLA claims database for April 1995 to February 2006, to identify all claims apparently relating to falls. Claims were excluded from further analysis if, on scrutiny, they had not resulted from falls, or if they were still "open" (ie, unresolved). From the narrative descriptions of closed claims (ie, those for which the financial outcome was known), we developed categories of "principal" and "secondary" injury/harm and "principal" and "contributory" circumstance of falls. For each category, it was determined whether cases had resulted in payment and what total payments (damages and costs) were awarded. The proportions of contribution-specific injuries or circumstances to the number of cases and to the overall costs incurred were compared in order to identify circumstances that tend to be more costly. Means were compared and tested through analysis of variance (ANOVA). The association between categorical variables was tested using the chi-square test. RESULTS: Of 668 claims identified by word search, 646 met inclusion criteria. The results presented are for the 479 of these that were "closed" at the time of the census. Of these, 290 (60.5%) had resulted in payment of costs or damages, with the overall total payment being 6,200,737 pound (mean payment 12,945 pound). All claims were settled out of court, so no legal rulings on establishing liability or causation of injury are available. "Falls whilst walking;" "from beds or trolleys" ("with and without bedrails applied") or "transferring/from a chair" were the most frequent source of these claims (n = 308, 64.2%). Clear secondary contributory circumstances were identified in 190 (39.7%) of closed claims. The most common circumstances cited were "perioperative/procedural incidents" (60, 12.5%) and "requests for bedrails being ignored" (54, 11.3%). For primary injuries, "hip/femoral/pelvic fracture" accounted for 203 (42.4%) of closed claims with total payments of 3,228,781 pound (52.1% of all payments), with a mean payment 15,905 pound per closed case. A "secondary" contributory circumstance could be attributed in 133 (27.8%) of cases. Of these, "delay in diagnosis of injury," "recurrent falls during admission" and "fatalities relating to falls" were the commonest circumstances (n = 59, 12.2%). DISCUSSION: Although falls are the highest volume patient safety incident reported in hospital trusts in England, they result in a relatively small number of negligence claims and receive a relatively low total payment (0.019% in both cases). The mean payment in closed claims is also relatively small. This may reflect the high average age of the people who fall and difficulty in establishing causation, especially where individuals are already frail when they fall. The patterns of claims and the narrative descriptions provide wider lessons for improving risk-management strategies. However, the inherent limitations and biases in the data routinely recorded for legal purposes suggest that for more informative research or actuarial claims analysis, more comprehensive and systematic data to be recorded for each incident claim are needed.