Increased serum concentration of procalcitonin (PCT) in limited number of paediatric patients with acute severe bacterial infections has been described previously. In a prospective study in 337 hospitalised adult patients fulfilling the SIRS-criteria, serum PCT was determined on admission and up to 9 days thereafter. Patients with microbiologically documented infection showed peak values of 30 ng/ml at day 3, which rapidly decreased to normal levels. Patients without sepsis revealed baseline values (0.1 ng/ml or lower). The validity criteria were calculated for several breakpoints of PCT. We detected an interval from 0.1 to 0.5 ng/ml under which a severe microbial infection is unlikely (sensitivity 91%, specificity 25%, positive predictive value 39%, negative predictive value 86%). An infection is most likely above 0.5 ng/ml (sensitivity 60%, specificity 79%, positive predictive value 61%, negative predictive value 78%). Between these two points an infection can neither be confirmed nor excluded. The excellent specificity and negative predictive value at a cut-off point of 0.5 ng/ml suggests that this test might be a useful parameter in the management of infectious diseases.