Algorithm for enrollment and randomisation. Endotracheal intubated neurological or neurosurgical patients were routinely assessed during morning and evening rounds for eligibility criteria. Consent was obtained from patients' surrogates in medically and neurologically stable patients without anticipated neurological deterioration. Consented patients subsequently underwent a 30 minute T-piece trial and Airway Care Score (ACS) assessment. If the patient failed either assessment, they were re-evaluated in 12 hours. Patients that passed both tests were randomised to early or delayed extubation. Patients randomised to delayed extubation had their Glasgow Coma Score (GCS) reassessed at least every 12 hours. If the GCS improved to more than 8 and they passed the above T-piece and airway reassessments, they were immediately extubated. If the patients neurological status did not improve after several assessments a trial of extubation could still be considered at the discretion of the attending physician to avoid the necessity of placing a tracheostomy.