Abstract. Objective: In mechanically ventilated patients flow limitation often goes unrecognised. We compared three methods for detection of flow limitation in mechanically ventilated patients: the resistance method, the negative expiratory pressure (NEP) method, and the interrupter method. Design: Prospective study. Setting: A medical intensive care unit in a university hospital. Patients: Twenty-six patients (20 COPD, six other pathology), mechanically ventilated under sedation. Measurements and Results: Respiratory mechanics were obtained during application of the three methods. For the resistance method, flow limitation was expressed as percentage of expiratory tidal volume, in which flow did not decrease (FLP-R). For the NEP method, flow limitation was expressed as percentage of expiratory tidal volume in which flow did not increase (FLP-NEP). For the interrupter method, flow limitation was expressed as area of spike-overshoot in flow after interruption. In 18 COPD patients, flow limitation was detected by all methods; mean FLP-R 76% (SD 12%), mean FLP-NEP 90% (SD 11%), mean spike area 21 ml (SD 7 ml). In three patients with other pathology, these values were, respectively, 20% (SD 19%), 48% (SD 21%), and 5 ml (SD 4 ml). The three methods were in close agreement. In nine patients the resistance method increased flow and in six patients the NEP method decreased flow compared to the unimpeded breath. Conclusions: In mechanically ventilated patients, flow limitation can well be detected by the resistance-, NEP-, and interrupter methods. However, the NEP method can overestimate the flow limited portion, while the resistance method can underestimate the flow limited portion. The interrupter method is found to be less practical.