The manner in which elements of clinical history, physical examination and investigations influence subjectively assessed illness severity and outcome prediction is poorly understood. This study investigates the relationship between clinician and objectively assessed illness severity and the factors influencing clinician's diagnostic confidence and illness severity rating for ventilated patients with suspected pneumonia in the intensive care unit (ICU). A prospective study of fourteen ICUs included all ventilated admissions with a clinical diagnosis of pneumonia. Data collection included pneumonia type - community-acquired (CAP), hospital-acquired (HAP) and ventilator-associated (VAP), clinician determined illness severity (CDIS), diagnostic methods, clinical diagnostic confidence (CDC), microbiological isolates and antibiotic use. For 476 episodes of pneumonia (48% CAP, 24% HAP, 28% VAP), CDC was greatest for CAP (64% CAP, 50% HAP and 49% VAP, P<0.01) or when pneumonia was considered "life-threatening" (84% high CDC, 13% medium CDC and 3% low CDC, P<0. 001). "Life-threatening" pneumonia was predicted by worsening gas exchange (OR 4.8, CI 95% 2.3-10.2, P<0.001), clinical signs of consolidation (OR 2.0, CI 95% 1.2-3.2, P<0.01) and the Sepsis-Related Organ Failure Assessment (SOFA) Score (OR 1.1, CI 95% 1.1-1.2, P<0.001). Diagnostic confidence increased with CDIS (OR 16.3, CI 95% 8.4-31.4, P<0.001), definite pathogen isolation (OR 3.3, CI 95% 2.0-5.6) and clinical signs of consolidation (OR 2.1, CI 95% 1.3-3.3, P=0.001). Although the CDIS, SOFA Score and the Simplified Acute Physiologic Score (SAPS II) were all associated with mortality, the SAPS II Score was the best predictor of mortality (P = 0. 02). Diagnostic confidence for pneumonia is moderate but increases with more classical presentations. A small set of clinical parameters influence subjective assessment. Objective assessment using SAPS II Scoring is a better predictor of mortality.