A prospective, observational study was conducted in a medico-surgical intensive care unit to assess the value of C-reactive protein (CRP), temperature and white cell count (WCC) measurements for the diagnosis of infection in critically ill patients. CRP, temperature and WCC were monitored daily in 76 infected and 36 non-infected patients. Multiple receiver-operating characteristics (ROC) curves were used to compare each parameter for infection diagnosis. The area under the curve (AUC) of CRP was significantly higher than that of temperature (0.93 and 0.75, respectively; p < 0.001). A CRP concentration of >8.7 mg/dL and a temperature of >38.2 degrees C were associated with infection, with a sensitivity of 93.4% and 54.8%, and a specificity of 86.1% and 88.9%, respectively. The ROC curve of WCC showed a poor diagnostic performance. The combination of CRP and temperature increased the specificity for infection diagnosis to 100%. In the subgroup of patients with ventilator-associated pneumonia (n = 48), CRP measurements were more reliable than temperature (AUC 0.92 and 0.78, respectively; p 0.006). The CRP levels in infected patients with sepsis, severe sepsis and septic shock were 15.2 +/- 8.2, 20.3 +/- 10.9 and 23.3 +/- 8.7 mg/dL, respectively (p 0.044). It was concluded that CRP was a better marker of infection than temperature. However, the combination of CRP and temperature measurements further increased the specificity for infection diagnosis, even in the subgroup of patients with VAP.