Our objective was to estimate the 5-year survival to death in insured warmblooded riding horses (as reflected by a life-insurance claim) in horses with or without at least one costly veterinary-care event (VCE) in 1997. We also determined the impacts of gender, age, previous diagnosis of veterinary claim, geographic variables, cost for veterinary claims and life-insurance value upon the survival. The design was a retrospective cohort study using a population of warmblooded riding horses insured in a Swedish animal-insurance company (Agria Insurance, P.O. 70306, SE-107 23 Stockholm, Sweden) in 1997. All horses in this population had insurance both for veterinary-care and life ("complete" insurance reimbursing the insurance holder for most medical problems). We followed two cohorts (2,495 horses with > or = 1 VCE and 15,576 horses with no VCE in year 1997). The median VCE cost in horses with > or =1 VCE was 3,800 SEK, with 10th and 90th percentiles of 1,400 and 11,400 SEK respectively. In total 944 of the 2,495 horses (38%) in the VCE-positive cohort and 2,962 of the 15,576 horses (19%) in the VCE-negative cohort had died (had a settled life-insurance claim) after the follow-up time. An exponential-regression model showed that geldings had a higher risk of claimed death compared to mares and mares had a higher risk compared to stallions. The risk of death increased linearly with age. The risk of death increased with increasing life-insurance value. Horses with previous lameness had the lowest survival, and for high-cost claims this increased risk was not associated with age.