Background/Aims: Delirium is a common and serious complication in hospitalised patients and its pathophysiology is incompletely understood. We aimed to examine whether blood-cerebrospinal fluid barrier dysfunction, as measured by Q-albumin (the ratio of cerebrospinal fluid albumin to serum albumin), was associated with delirium. Methods: In this prospective cohort study of hip fracture patients from Oslo University Hospital, Norway, serum was collected preoperatively and cerebrospinal fluid just before the onset of spinal anaesthesia. Albumin levels in serum and cerebrospinal fluid were analysed consecutively, and Q-albumin was calculated using the formula [cerebrospinal fluid albumin (mg/dl) × 1,000]/[serum albumin (mg/dl)]. Q-albumin >10.2 was used as the cut-off for blood-cerebrospinal fluid barrier dysfunction. Patients were assessed daily for delirium using the Confusion Assessment Method. Results: Out of 120 patients, 69 had delirium, 22 had subsyndromal delirium, and 29 were free from delirium. The majority of patients, i.e. 106 (88%), had intact blood-cerebrospinal fluid barrier integrity, but all 14 patients with blood-cerebrospinal barrier dysfunction had delirium (n = 11) or subsyndromal delirium (n = 3). Conclusions: The results suggest that blood-cerebrospinal fluid barrier dysfunction may be relevant for delirium pathophysiology when it occurs. However, the low prevalence (16% of delirium patients) indicates that this is not a prerequisite for the development of delirium.