Summary Objectives Identification of infection remains a major challenge, particularly in acutely ill patients. The Infection Probability Score (IPS) was developed to help rule out infection in acutely ill patients. In the present study, we determined the IPS in acutely ill, intensive care unit (ICU) patients to assess its use in the diagnosis and treatment of infection. Methods In this prospective, observational study, we enrolled 107 consecutive patients who were admitted to the ICU without antibiotic therapy. Patients were allocated to four groups according to the probability of infection determined from clinical and microbiological data and their IPS values were then evaluated daily throughout the ICU stay. Results The IPS was higher in patients with the highest clinical probability of infection and decreased significantly in these patients after 5 days of effective antimicrobial therapy. The IPS remained below the cut-off value in non-infected patients. Patients in whom inadequate antimicrobial therapy was administered had a greater mortality than the other patients. Conclusions The IPS had a good predictive value for diagnosis of infection. In addition, dynamic evaluation of this score may help to assess the response to therapy.