The objective was to improve the ability of general practitioners (GPs) to diagnose depression and dementia compared with standard screening measures. The setting was a retirement village on the outskirts of Sydney, Australia. The study used a prepost design with a 6 month follow-up. The intervention involved a visit to the GP by an academic detailer who spent 15 minutes discussing the diagnosis of depression and dementia. Ratings of depression and dementia on two occasions by GPs, and by independent interviews were made using the Geriatric Depression Scale, Mini-mental State Examination and Canberra Interview for the Elderly. In the case of depression, the level of agreement (Kappa) between the GPs and all instruments increased significantly by a factor of between 2.3 and 3.3. The doctors did not significantly improve in their agreement with the instruments on the diagnosis of dementia. An academic detailing approach to improving GPs' abilities in the diagnosis of depression can be effective. A controlled trial would be justified to confirm this finding.