The lack of social support is the most common risk factor for late-life suicide. Few previous community-based interventions against the lack of social relationships reduced suicide. This study aims to evaluate outcomes of a community-based prevention program against suicide amongst the elderly in rural Japan. During 1995-2002, the program based on population strategy including group activity, psychoeducation and self-assessment of depression but no screening for depression, was implemented for elderly residents in Yuri town, Japan (5-year average population 6817; 5-year average suicide rate [> or =65 years old] 291.4/10(5)). Changes in the relative risk of suicide for individuals (> or =65 years old) before and after the 8-year implementation were estimated by the incidence rate ratio (IRR), using a quasi-experimental design with a neighboring reference, Chokai town, Japan (5-year average population 8136; 5-year average suicide rate [> or =65 years old] 216.5/10(5)). The risk of elderly females in Yuri completing suicide was reduced by 76% (age-adjusted IRR, 0.24; 95% CI, 0.10-0.58), while there was no change in the risks for Yuri's elderly males and both Chokai's elderly males and females. General loglinear analysis estimated a ratio of the female IRR in Yuri to that in its Akita prefecture of 0.35 (95% CI, 0.14-0.84), showing that the reduction of the risk in the intervention area was greater than a historical trend. A community-based suicide prevention through a group intervention designed to increase knowledge and to cultivate social relationships would be effective for elderly females but not males.