AIM: Using the unique Danish psychiatric and somatic health registers, we investigated the rate of subsequent dementia in patients with late-onset acute and transient psychosis. METHODS: By linkage of the psychiatric and the somatic nationwide registers of all patients with in- or outpatient hospital contact in Denmark, we included all patients with a first ever contact during the period 1 January 1994 to 31 December 2001 with one of the main index diagnoses: late-onset acute and transient psychosis or osteoarthritis. Data on the general population were also included. The first diagnosis of dementia for each individual at discharge or at outpatient contact was established. Poisson regression models were used to compare the cohorts of patients with dementia as the outcome of interest. RESULTS: Using a cut-off age of 60 years, 8062 individuals were included. Significant associations were found between a subsequent diagnosis of dementia and the index diagnosis, age and calendar time. Overall, the rate ratio for developing dementia in late-onset acute and transient psychosis compared to osteoarthritis patients was 10.86 (95% confidence intervals, 8.42 and 14.00, respectively), however, the magnitude of the rate ratio varied according to sex, age, duration since diagnosis and calendar time. Compared to the general population, the rate ratio was 8.12 (95% confidence intervals, 6.77 and 9.74, respectively). CONCLUSION: The present study has established that subjects with late-onset acute and transient psychosis are at 11 times higher risk of subsequently getting a diagnosis of dementia compared to patients with osteoarthritis, and at 8 times higher risk compared to the general population.