Greater breath-to-breath variability of breathing pattern during a spontaneous breathing trial (SBT) has been reported to indicate adequate respiratory reserve. A coefficient of variation (CV) calculation of certain tidal ventilation parameters has been suggested to reliably index variability, with reduced parameter variability indicating inability to wean from mechanical ventilation. We hypothesized that breathing pattern variability would differ before and during the withdrawal of fully supported ventilation with pressure support (PS). Methods: Our electronic medical record system allows interrogation of monitored values at one minute intervals. Ventilator data was captured every minute for one hour as sedation was reduced prior to and during the reduced pressure support period (SBT) that preceded the decision to extubate or continue ventilation. Reasons for failure-to-wean were retrieved from the medical record. Fifteen unselected medical and surgical ventilated patients were monitored in 19 consecutive weaning periods; 5 patients were extubated, while 14 weaning trials led to a primary care giver’s decision to continue fully supported ventilation. Results: Parameter variability was similar before and during SBT in all patients. Against expectations, CV variability of the rapid shallow breathing index (RSBI = f/Vt) was greater in the failed trials than in those where SBT was well tolerated. Reasons recorded for failure-to-pass the SBT included inattentiveness, increased RSBI, excessive secretions and weakness. Conclusions: 1) In unselected patients, variability of commonly monitored ventilatory parameters may be similar before and during the SBT. 2) The RSBI often varies markedly during the SBT, particularly in those destined to fail. Selection criteria must be applied before CV-assessed breathing pattern variation can be used reliably as a clinical tool to predict breathing reserve.