We considered whether general practitioners should examine all older patients over a certain age for cognitive impairment in screening for early dementia. We invited presentations from key experts, selectively reviewed the literature, and developed a consensus statement. The efficacy of and benefits from unselective use of cognitive testing and informant questionnaires for detecting early dementia in older patients attending general practice are limited. Positive predictive values of cognitive screening for dementia are less than 50%, even for older patient populations. Higher values may be obtained by testing patients who have a relevant history of cognitive or functional decline. Whatever procedures are adopted for screening older general practice attenders for cognitive impairment or early dementia, investigation is still required into the relative merits of different health professionals performing the screening, the positive and negative effects on patients and their families, and the cost-benefit ratio. The majority view of workshop participants was that cognitive testing should occur for older patients when there is a reason to suspect dementia. Testing may occur in an individual considered to be at risk because of an informant history of cognitive or functional decline, clinical observation, or, sometimes, very old age. No single instrument for cognitive screening is suitable for global use. Screening programs must be supported by training and supplemented by education for professionals and families in management of dementia.