Purpose To evaluate if preoperative assessment with A Quick Test of Cognitive Speed (AQT) could increase the accuracy of predicting delirium after cardiac surgery compared to Mini-Mental State Examination (MMSE), and examine if a composite of variables, including cognitive function and depressive symptoms, could be useful to predict delirium. Patients and Methods Cardiac surgery was performed in 218 patients (mean age 72 years). Preoperative evaluation involved AQT, MMSE and Hospital Anxiety And Depression Scale (HADS). Postoperative delirium was assessed using Nursing Delirium Screening Scale (Nu-DESC) and Confusion Assessment Method-ICU (CAM-ICU). Logistic regression was performed to detect predictors of postoperative delirium and receiver operator characteristic curves (ROC) with area under the curve (AUC) to determine the accuracy. Results Postoperative delirium occurred in 47 patients (22%) who had lower MMSE scores (median (range), 27 (19–30) vs 28 (20–30), p =0.009) and slower AQT (median (range), 76 (48–181) vs 70 (40–182) seconds, p =0.030) than patients without delirium. Predictive power measured as AUC (95% CI) was 0.605 (0.51–0.70) for AQT and 0.623 (0.53–0.72) for MMSE. Logistic regression (OR, 95% CI) showed MMSE <27 points (2.72, 1.27–5.86), AQT >70 sec (2.26, 1.03–4.95), HADS-D >4 points (2.60, 1.21–5.58) and longer cardiopulmonary bypass-time (1.007, 1.002–1.013) to be associated with postoperative delirium. Combining these parameters yielded an AUC of 0.736 (0.65–0.82). Conclusion The ability of predicting delirium using AQT was similar to MMSE, and only slightly higher by combining AQT and MMSE. Adding HADS-D and cardiopulmonary bypass-time to MMSE and AQT increased the predictive power to a borderline acceptable discriminatory value. Preoperative cognitive tests and screening for depressive symptoms may help identify patients at risk of postoperative delirium. Yet, there is still a need to establish useful preoperative tests.