Neuropsychological impairment following cardiopulmonary bypass (CPB) remains a serious consequence of otherwise successful surgery. The incidence of neuropsychological decline is poorly understood due to varied measurement intervals, and perhaps more importantly the use of unreliable detection and classification methods. The reported incidence varies considerably, ranging anywhere from 30% to 90% of subjects. While the nature of this impairment has not been fully elucidated, recent evidence suggests that microembolism during surgery may be the principal causative agent of postoperative cerebral dysfunction. The work described in this thesis investigates one possible source of microembolism leading to postoperative decline, namely thromboembolism arising from excessive activation of the haemostatic mechanism. Crucial to the accurate detection of significant decline in individual patients, this work also focuses on the development and use of meaningful criteria to be used when describing change in neuropsychological performance measures. The strong haemostatic activation during CPB is controlled by heparin anticoagulation. The clinical performance of the Hepcon heparin-monitoring instrument was compared to the activated clotting time (ACT), which is used in most cardiac centres. An analysis of samples from 42 elective coronary artery bypass grafting (CABG) patients shows that the ACT does not detect the significant decline in heparin concentration seen upon connection to CPB, in comparison to the Hepcon. The Hepcon appears to be in satisfactory agreement with laboratory anti-Xa analysis of heparin concentration, with the mean difference for the Hepcon at -0.46 U/ml, and the limits of agreement +/- 1.12 U/ml. Further analysis shows that that for 95% of cases, the Hepcon will give values that are between 0.53 and 1.27 times the value for anti-Xa. The loss of relationship between ACT and heparin concentration was further investigated by converting ACT values to heparin concentration. The results provide data on the degree of prolongation in ACT times brought about by factors associated with CPB. A methodology is presented by which users can adjust for the loss of relationship between ACT and heparin. This work also demonstrates that under normal usage of the ACT, the user may obtain values up to 3 times appropriate for the plasma heparin concentration. The computer-administered neuropsychological testing tool (the MicroCog) was validated using 40 age-matched control subjects. Using a two-week interval, the summary score correlation coefficients ranged from .49 to .84, with all scores demonstrating significant practice effects. Also presented are retest normative data that may be used to determine significant change in a homogeneous sample using both reliable change and regression models of analysis. The performance of four different models of change analysis was then analysed using data from the clinical group. The regression technique of analysis was shown to be the most useful prediction model as it provides correction for both practice effects and regression toward the mean in each individual. A novel statistical rationale is presented for the choice of criteria in the identification of patients that may be defined as overall impaired when using a battery of test scores. When using one-tailed prediction models for decline, the binomial distribution of scores was shown to be a useful descriptive statistic providing an estimate of change due to chance. When applied to a suitable selection of scores that minimise shared variance, a value +/- 20% of test scores used was demonstrated to be a rational cut-off for an individual to be classified as impaired. Using this methodology, 32.7% of patients were identified as significantly deteriorated in neuropsychological test function immediately prior to discharge from hospital. Patient age was shown to be a significant predictor of neuropsychological decline following CPB. No significant relationship was identified between thrombin generation and neuropsychological change scores, however problems with patient recruitment and retention limited the statistical power of this study. An intriguing relationship with heparin concentration was noted that might warrant further investigation. This work highlights the complex nature of post-bypass neuropsychological dysfunction and the complexities in assessing decline. The regression-based model was shown to be highly useful in the analysis of data from a suitably validated neuropsychological testing tool. The argument that no suitable criterion exists for the identification of patients as overall impaired has been challenged with the development of a rational cut-off based on the likely distribution of change scores across a series. The work presented here confirms the need for standardised testing methods based on sound statistical criteria. This work also highlights the problems associated with current methods for monitoring anticoagulation therapy during bypass surgery. Methodology is presented that allows adjustment of ACT results to account for CPB-induced prolongation of clotting times. Current techniques for heparin monitoring overestimate heparin levels on bypass by up to threefold, which may predispose to subclinical coagulation and increased delivery of protamine.