This dissertation studies factors that predict use of public eldercare, informal care, and purchase of private services in relation to an individual’s needs, social network characteristics, and sociodemographic factors. A further purpose is to examine whether use of public eldercare is correlated to receipt of informal care and purchase of private services in the Swedish welfare state. The dissertation is based on the Kungsholmen Study, a population-based longitudinal study. Studies I–III used cross-sectional data from community-dwelling people aged 81-100 and examined (I) gender, (II) marital and parental status, and (III) dementia and depressive symptoms as predictors of use of home help. Study IV analyzed factors related to moving into institutional care and receipt of home help from 1994/96 to 2000. The majority of support provided to elders living in the community comes from informal sources, even among people living alone. There was considerable overlap between home help and informal care. When all sources of care were considered, childless individuals had comparatively lower odds of receiving care. Factors predicting use of public eldercare and informal care differed depending on whether or not elders coresided. No gender differences in use of formal and informal care were found when controlling for household composition. Living alone, dementia, need of help with household chores, and walking limitations increased the likelihood of using public eldercare. Coresidence, informal care from outside the household, and use of private services decreased the likelihood. Depressive symptoms increased the likelihood of receiving home help and institutionalization when using longitudinal data, but not in the cross-sectional studies. Educational level was of importance and interacted with several factors; persons with higher levels of education were advantaged. Very few people moved into institutional care without previously having received home help services. Essentially the same factors that predicted receipt of home help services also predicted institutionalization.