Background: Respiratory reserve is a vital determinant of the ability to successfully discontinue ventilatory support. The coefficient of variation (CV) of routinely measured respiratory variables may be one convenient indicator of readiness to be weaned. We examined the CV of respiratory rate (RR), minute ventilation (MV), tidal volume (Vt), and rapid shallow breathing index (RSBI) both prior to and during partial ventilatory support to evaluate the discriminatory value of variability in predicting weaning success. Methods: A serial data logger continuously streamed RR, MV, and Vt during controlled mechanical ventilation for the eight hours preceding and during a series of pressure support (PS) reduction trials (10 minutes each) in a mixed surgical and medical ICU population. Data for each selected pressure support level (12 à 10 à 8 à 7 à 5) was obtained until the patient was successfully extubated or failed the weaning attempt. Results: Thirty two mechanically ventilated patients judged to be appropriate for attempting discontinuation of mechanical ventilation within the next 24 hours were enrolled. Ten patients failed to extubate secondary to persistently altered mental status. For these patients, CV trends in respiratory parameters during the graded support reduction were unrevealing. Of the remaining 22 patients, 7 patients failed to extubate with the most common reason being tachypnea or low Vt. Fourteen patients were successfully extubated, of which only 1 required re-intubation within 48 hours. Increased and maintained variability of Vt during PS ventilation appears to discriminate successful from failed weaning trials (figure). Variability trends for RR and MV during controlled mechanical ventilation do not appear to be strong indicators of weaning success. During the graded PS ventilation, the variability of RR, MV, and RSBI also appear to be less useful as indications of weaning success. Conclusion: Although data collection is ongoing, increased and maintained variability during graded reduction in pressure support ventilation, assessed by coefficient of variation, appears to be a useful predictor of weaning success. Data for determination of CV is readily attainable from modern mechanical ventilators.