Background and aims: Existing information about institutionalization of elderly individuals is mainly based on cross-sectional data and does not address the cumulative risk of institutionalization. The purpose of the present study was to analyze longitudinal data prospectively and estimate the risk of placement in an elder care institution for individuals aged 70 years or older. Methods: The study was based on a longitudinal investigation (the H70 study) of a random sample of 70-year-olds living in Gothenburg, Sweden, in 1971. Individuals were followed from age 70-100 years. Three different analyses were performed: a descriptive prospective analysis, cross-sectional analyses at ages 70, 79 and 85 years, and a longitudinal analysis of predictors for institutionalization. Results: The prospective analysis indicated that 50% of the individuals eventually moved to an elder care institution. Significantly more women than men were institutionalized, although for women the move occurred later in life. Cross-sectional analyses demonstrated that various factors were important to institutionalization at different ages. The Cox regression model with time-varying covariates indicated that gender, socio-economic situation, marital status, number of symptoms, having children living nearby, and activities in daily life were related to institutionalization. Conclusions: The proportion of elderly persons relocating to institutions was significantly higher than that generally found in cross-sectional studies. It was possible to identify variables that predict institutionalization during a subsequent 30-year period, but different analyses revealed different effects from the factors evaluated.