OBJECTIVES: To determine the annual incidence of severe sepsis in the Netherlands by performing a point-prevalence survey in multiple intensive care units. METHODS: ICU's were invited to participate in a one-day survey and monitor patients during the first 24 hours of their stay, if they were admitted with a proven/suspected infection. Patient-specific questionnaires captured demographic and clinical information, presence of Systemic Inflammatory Response Syndrome (SIRS), and the functional status of seven organ systems. The annual national incidence was calculated from the results of the survey following two approaches: 1) by multiplying the survey incidence (patients/day) with the number of days per year, corrected for the fact that the survey was held on a weekday, and 2) by using the relation P=I·D (P for prevalence, I for incidence and D for duration of stay) in which P was known from the survey and D was estimated as the geometrical mean of duration-to-date. RESULTS: Forty-seven ICU's participated (in 43 general and 4 university hospitals), representing 42% of the national admission capacity. During the study-period, 18 patients meeting criteria for severe sepsis were newly admitted, and another 116 patients with severe sepsis were already present. According to the first calculation method, the annual incidence of severe sepsis was 13,137 ± 2,821 patients, whereas the second method, with an estimated D of 13.3 ± 1.1 days, lead to a calculated incidence of 8,643 ± 929 patients/year. CONCLUSIONS: Using the results from a point-prevalence survey, different approaches lead to different outcomes. Both approaches hold advantages and disadvantages. The second method is considered superior because it is based on a larger population and is less sensitive to daily variations.