It is important to know when improvement or deterioration in cognitive function occurs. Until fairly recently neuropsychologists have made these judgments clinically, with little resort to empirical methods. In addition to the issue of whether a change in performance is reliable, there is also the consideration of whether it is clinically significant. This article briefly discusses these concepts, reviews the most common methods for determining reliability in change in test scores, considers their use in the broader clinical context, and illustrates their application with reference to an actual database.