OBJECTIVE: To determine the demographic and health factors related to the frailty syndrome in older adults. METHODS: This is a longitudinal quantitative study carried out with 262 older adults aged 65 years and older, of both sexes, living at home. Data collection was carried out in Period 1 between October 2007 and February 2008, and in Period 2 between July and December 2013. For data collection, we used the sociodemographic profile instrument, the Edmonton Frail Scale, the Mini-Mental State Examination, the number of falls in the last 12 months, the number of self-reported diseases and used drugs, the Functional Independence Measure, and the Lawton and Brody Scale. We used descriptive statistics for data analysis, in the comparison of the means between periods, the nonparametric Wilcoxon test, and the method of Generalized Estimating Equations, which is considered an extension of the Generalized Linear Models with p = 0.05. RESULTS: Of the 515 participants, 262 completed the follow-up, with a predominance of females, older individuals, and those who had no partner; there was an increase in frail older adults. In the Generalized Estimating Equations analysis, frailty score was related to sociodemographic (increase in age, no partner, and low education level) and health variables (more diseases, drugs, falls, and decrease in functional capacity). There was an association between the variables of age (older), marital status (no partner), and loss of functional capacity. CONCLUSIONS: Frailty syndrome was associated with increasing age, having no partner, and decreased functional capacity over time, and investments are required to prevent this syndrome and promote quality in aging.