Assessment of cognitive change in individual patients may be confounded by unreliability of test scores and effects of repeated testing. An index correcting for both problems is proposed and compared with change indices that do not or do not adequately deal with measurement error and practice effects. These indices were used to examine cognitive deterioration in a sample of 63 patients undergoing cardiac surgery. It was demonstrated that for test measures with a low reliability, failure to correct for measurement error resulted in overestimation of deterioration rates. For test measures with a high reliability, but showing substantial practice effects, failure to correct for practice effects resulted in underestimation of deterioration rates. With the proposed index, cognitive deterioration shortly after cardiac surgery was most frequently observed for attention and psychomotor speed, less frequently for verbal fluency, and only occasionally for learning and memory.