In this issue of CHEST (see page 1197), MacCallum et al7 make us rethink the predictive value of SIRS, using a large cohort of patients in a single ICU who had undergone cardiac surgery. Patient characteristics, organ dysfunction, outcomes, and detailed data on the number of SIRS criteria met during each hour from ICU admission to death or discharge were prospectively collected. They found the traditional definition of two or more SIRS criteria was met by almost 60% of patients in the first hour after surgery and by > 95% within the first day. The low positive predictive value (PPV) for mortality associated with meeting two or more SIRS criteria in the first 24 h (PPV, 2.8%) was improved by modifying the definition to meeting three or more or four SIRS criteria (PPV, 4.2% and 10.2%, respectively), or to fulfilling two or more criteria for at least six consecutive hours on the first day (PPV range, 5.4%-7.0%). At first, these low predictive values for mortality might suggest SIRS criteria are not very useful, until one considers that cardiac surgery is rarely fatal, and essentially all patients with poor outcomes could be identified on the first ICU day by applying these reasonably simple modifications of the original SIRS criteria.