The Location Learning Test (LLT) is a brief, new measure of visuo- spatial learning that has been developed for use with older adults and in dementia. It does not require fine motor control, verbal responses, or complex instructions. The validity of the LLT was established by comparing the performance of three groups of subjects: normal elderly controls, patients with Alzheimer's disease and patients with vascular dementia. There were significant differences between normal subjects and those with dementia, including those with mild dementia (MMSE ≤ 20). Performance on the LLT was not predicted by premorbid IQ or age, but did correlate highly with the MMSE (R = .77). A cut-off score was selected which yielded a sensitivity of 100%, specificity of 82.8%, and a positive predictive value of 83.3%. Two groups of dementia patients were found; those who were able to improve their performance through learning and those who were not. The ability to learn was not predicted by degree of cognitive impairment as measured by the MMSE. Female subjects with dementia performed significantly worse than male subjects with dementia and this effect was not a consequence of level of impairment as measured by MMSE, premorbid IQ (NART), or diagnosis. Normative data is currently being collected.