Background. A method of defining change in neuropsychologic test scores that accounts for test reliability and practice effects was applied to determine accurately the incidence of acquired neuropsychologic deficits after coronary artery bypass grafting. Methods. Neuropsychologic assessment was performed on 50 patients before and at 7 days after either hypothermic or normothermic coronary artery bypass grafting. From a matched control group of 24 normal subjects who were examined twice over a similar interval, reliable change indices that controlled for measurement error and practice effects were calculated for each neuropsychologic measure. With the use of these indices, the incidence of postoperative decline among the study patients was determined. For comparison, the incidence of decline using the 'one standard deviation' criterion also was calculated. Results. Comparing the reliable change and standard deviation methods, statistically significant differences in the incidence of decline were observed in 5 of 11 neuropsychologic measures. The reliable change method identified more patients with neuropsychologic deficits on most measures. Conclusions. The control of measurement error and practice effects can alter significantly the calculated incidence of neuropsychologic impairment after coronary artery bypass grafting.