Critical Flicker Fusion Threshold (CFFT) is a psychophysical threshold and, in psychological terms, it is regarded as a measure of information processing capacity. CFFT has previously been shown to be a valid and reliable measure in young healthy volunteers and it also has a long history of use as a psychopharmacological measure in this group. Furthermore, the test satisfies many of the requirements of an 'ideal' measure for monitoring change, especially in a psychopharmacological context. Despite this, CFFT has been neglected as a research tool in elderly and Alzheimer's disease (AD) populations and was therefore investigated further in this regard. CFFT in community-based healthy elderly subjects was normally distributed, but CFFT and ascending and descending thresholds were not significantly correlated with age. The difference between ascending and descending thresholds was, however, significantly correlated with age and this relationship appeared to be due almost entirely to a change in the descending threshold. In addition, descending thresholds were found to be significantly greater than ascending thresholds in healthy elderly subjects. In contrast, patients with AD were found to have significantly lower CFFT and descending scores compared with healthy elderly subjects. Interestingly, descending thresholds were significantly lower than ascending thresholds in the patient group, a feature that may be a characteristic of AD. Mean CFFT and ascending and descending thresholds were found to have a high test-retest, split-half and inter-rater reliability, in addition to being significantly correlated with a number of psychometric measures, clinical scales and neuropsychological instruments commonly used to assess patients with AD. CFFT is a quick and simple measure to administer and patients had no difficulty completing the test. Because the measure is a psychophysical threshold, it is free from educational and cultural bias and there are no floor or ceiling effects. From the results of this work, CFFT appears to be a useful research tool in AD. It may be a suitable measure for monitoring cognitive change over time, either in community studies of AD or a clinical trial context, but further work is required. The technique might also contribute to the early detection of AD. This application would be particularly important because it would enable effective pharmacotherapies to be started early during the course of the illness before neuronal damage is too advanced and this would have significant benefits for patients. Copyright 2000 John Wiley & Sons, Ltd.