Summary Despite the importance of healthcare-associated infection, few studies have quantified the association between severe infection and hospital exposure in UK populations. Our aim was to estimate the proportion of the population with recent hospital admission, together with rates of infection in hospital-exposed and hospital-naïve populations. We studied bacteraemia as a marker of severe infection in a population of 550 000, served by two hospitals, between 1 April 2000 and 31 March 2005. Hospital-exposed persons accounted for 8.3% of the population, defined as having been resident in a hospital in the last year. The hospital-exposed population accounted for 55% of all admissions, and 42% of emergency admissions to medical, paediatric or surgery departments. After adjustment for age, the hospital-exposed group had much higher rates of admission bacteraemia. Age-standardised incidence rate ratios relative to hospital-naïve patients were 43 95% confidence interval (CI): 22–85 for meticillin-resistant Staphylococcus aureus (MRSA), 20 (15–27) for S. aureus other than MRSA, 7.3 (5.2–10) for Streptococcus pneumoniae , and 14 (11–18) for E. coli . MRSA was common among hospital-exposed admissions, including emergencies in hospital-exposed men, rates of admission MRSA bacteraemia (31 per 100 000 per annum) and S. pneumoniae bacteraemia (33 per 100 000 per annum) were similar. This quantitative analysis confirms that prior hospital admission is a major risk factor for bacteraemia on hospital admission; it is unclear whether acquisition of pathogens in hospital, co-morbidity or other factors explain this.