A study was performed to estimate the prevalence of nosocomial infections (NI) and assess differences between medical care settings in one hospital complex. A seven-day period-prevalence survey was conducted in May 1998 in a large primary and tertiary healthcare centre in Geneva, Switzerland, that included all patients in acute, sub-acute and chronic care settings. Variables included demography, exposure to invasive devices and antibiotics, surgical history, and patients' localization. Overall prevalence of NI was 11.3% (acute, 8.4%; sub-acute, 11.4%; chronic care setting, 16.4%) in the 1928 patients studied, and ranged from 0% in ophthalmology to 23% in critical care units. Odds of infection in sub-acute and chronic care settings were significantly higher than in the acute care setting even after adjustment for case-mix [OR, 2.59; 95% confidence interval (CI(95)) 1.53-4.41; and OR, 2.34; Cl(95)1.38-3.95, respectively]. As a distinct group, patients in the geriatric location (belonging to the sub-acute care setting) showed a significant proportion of urinary (39%) and respiratory (21%) tract infections, contrasting with a relatively low exposure to urinary catheters (6.1%) and orotracheal intubation (0%). In conclusion, sub-acute and chronic care settings are associated with high infection prevalence even after case-mix adjustment. Prevalence studies are an easy surveillance tool that can be exploited further by analysing data according to hospital care settings to identify high-risk areas.