Relatively low use of nonsteroidal anti-inflammatory drugs (NSAIDs) and other analgesics has been noted in patients with probable Alzheimer disease (AD). Although this finding has been explained by a decline in patients' capacities to communicate about pain, self-report on pain of cognitively impaired elderly have been shown to be just as reliable as those of cognitively unimpaired elderly. However, previously published studies were aimed primarily at quantifying pain. Considering the various limbic areas affected in AD, a change also in the more qualitative, affective component of pain might be the cause of the low use of analgesics. Because affective disorders are highest in the early and middle stages of AD and decrease in the final stage, it was hypothesized in the present study that not only would the number of AD patients using analgesics would be lower than among a control group but, moreover, analgesic use would be lower in the early and middle stages of AD than in the final stage. The hypothesis was tested by comparing drug use (NSAIDs and analgesic non-NSAIDs) among 66 AD patients with that among 70 elderly people without dementia. The percentage of AD patients using analgesics was indeed significantly lower than among controls, but drug use was not dependent on the stage of AD. Consequently, our findings only partly support the hypothesis.