Background: Previously, key studies of the risk profile for post-surgical delirium have focused on general medical and non-elective patients, few have examined elective cohorts. Accurate prediction is imperative for clinical trials and prevention strategies. Aims/Hypothesis: Our hypothesis was that subtle pre-operative impairments of attention will be associated with risk of post-operative delirium. Method: A prospective study evaluating pre- and post-operative neuropsychological performance in older (≧70) consecutive elective admissions for orthopaedic surgery, and free of dementia (n = 100) was initiated in a general medical hospital. Results: Pre-operative attentional deficits were closely associated with delirium. Patients who developed post-surgical delirium had significantly slower mean reaction times (p ≤ 0.011) and greater variability of reaction time (p = 0.017). A 4- to 5-fold increased risk of delirium was observed for people one standard deviation above the sample means on these variables. Conclusions: The present study describes a measurement of attentional performance which could form the basis of a neuropsychological marker of delirium.