From 1998 to 2008, we performed a retrospective study in 289 nonimmunocompromised patients admitted to the ICU with community-acquired pneumonia and analyzed their diagnosis made according to microbiological methods. At least one bacteriological analysis at admission was performed in all patients; however, no viral diagnosis was performed at that time. In this study, 56% of patients had culture-negative pneumonia, and all patients received antibiotics according to the French guidelines. Mortality was 34% at day 28 for patients with microbiologically documented pneumonia and 11% for patients with nondocumented pneumonia (NDCAP). ICU length of stay was shorter for patients with NDCAP (P = .01). In patients with documented pneumonia, we performed a multiple correspondence analysis between admission parameters and a microbiological diagnosis to determine the pathogen for patients with NDCAP (Fig 1). Unfortunately, bacterial etiology could not be easily determined with this method. Thus, it seems difficult to assess outcome and readmission rate when almost one-half of the population could not be diagnosed by using microbiological techniques.