Aim Although numerous studies have reported that social participation (SP) has favorable effects on older people's health, no studies have focused on the autonomy of SP. We investigated the cross‐sectional relationship between the type, frequency, and autonomy of SP and self‐rated health (SRH). Methods A postal survey was mailed to 19 354 citizens born between 1945 and 1949 in a city in Nara prefecture (response rate: 65.9%). A total of 12 157 respondents (5114 men and 7043 women) with independent basic activities of daily living made up the analytical participants. Six types of SP were measured: volunteer groups, sports groups, hobby groups, senior citizens’ clubs, neighborhood community associations and cultural groups. Using multiple logistic regressions, we calculated the odds ratio for poor SRH, with non‐participation as the reference. Covariates included age, family size, body mass index, comorbidities, smoking, drinking, walking, depression and cognitive function. Results Adjusted odds ratios for poor SRH decreased with more frequent SP or with more voluntary SP among both sexes, but this dose–response relationship was stronger in women than in men. After mutual adjustment for the frequency and autonomy of SP, voluntary SP had stronger effects on SRH than frequent SP, especially in sports and hobby groups among men, and in volunteer and hobby groups among women. For cultural groups, women with frequent and voluntary SP had positive effects on SRH, but frequent and obligatory SP had negative effects on men. Conclusions The present results imply that encouraging voluntarily chosen SP rather than frequent SP might be a key point for successful aging. Geriatr Gerontol Int 2017; ••: ••–••.