This study investigated cognitive flexibility as indexed by the Written and Oral Trail Making Test (TMT) in sixteen persons with dementia and 60 normal elderly. Written and Oral TMT performances were significantly correlated with each other and with other tests of cognitive flexibility providing an index of the convergent construct validity of these tests. Part B of the Written TMT was influenced by psychomotor ability although this was diminished by the use of a ratio score B/A rather than simply Part B performance. The Oral test was not related to visual or motor skills, providing evidence for its divergent validity. Both parts of the Written and Oral TMT were found to be sensitive to cognitive decline in dementia. Performance on both Oral and Written TMT was also influenced by age, gender, education and intellectual ability. Previous norms (e.g., MOANS) for the Written TMT which are based on relatively educated elderly, tended to provide deflated scores for the community group tested here. Some additional, preliminary normative data for both tests were compiled which take into account the influences of each of these variables.