The study was undertaken in 1996 in the inner London borough of Tower Hamlets. All GPs working in the community were asked about their knowledge and experience of elder abuse; 73 GPs responded (68 percent response rate). The aspect of the study reported in this article involved asking respondents about 20 situations of suspected elder abuse. Overall, 84 percent of GPs knew of a patient in a situation in which it would be reasonable for them to suspect abuse. Seventy-five percent knew of a patient in a situation that involved a risk of psychological abuse, 52 percent a risk of neglect, and 47 percent a risk of physical abuse. The situations most commonly reported were those where it would be reasonable to suspect psychological abuse and possibly physical abuse; these situations invariably involved caregiving. Significant associations were found between the recognition of five or more risk situations and the age and qualifying date of the GP, their ethnic background (white GPs were more likely to recognize), the numbers of older people registered with the general practice, and the number of home visits paid by the GP. A significant association was found between the recognition of five or more risk situations and knowledge of an abuse case. These results raise important questions about whether GPs are failing to recognize abuse or are failing to define situations as abusive, as well as how they are responding to these situations. 7 tables, 1 figure, and 47 references