The diagnosis of ventilator-associated pneumonia (VAP) is problematic despite numerous attempts at defining acceptable diagnostic criteria and the optimal technique for routine respiratory sampling. Clinical criteria have imperfect diagnostic reliability in ventilated patients, but remain crucial for defining those patients who may require respiratory sampling. Quantitative clinical scoring systems may improve the accuracy of clinical diagnosis in some ventilated patients. Review of published studies suggest that fibreoptic bronchoscopic techniques have greater diagnostic reliability than qualitative endotracheal aspirates, despite inconsistent results when comparing the same techniques in different centres. However, the cost and invasive nature of bronchoscopic methods precludes their use as first-line techniques in VAP. Non-bronchoscopic, non-directed techniques are cheaper, safer and more widely available alternatives to fibreoptic bronchoscopy techniques and have comparable accuracy. Quantitation of respiratory tract cultures is useful in excluding VAP in patients with equivocal signs of pneumonia. The diagnostic threshold of bacterial load that defines the presence of VAP should vary according to the pre-test probability of pneumonia, length of ventilation, antibiotic administration and immunocompetence of the patient.