Background: The early identification of individuals at high risk for adverse outcomes by a Comprehensive Geriatric Assessment (CGA) in resource-limited primary care settings enables tailored treatments, however, the evidence concerning its benefits are still controversial. The main objective of this study was to examine the validity and reliability of the "Multidimensional Assessment of Older People in Primary Care (AMPI-AB)", a CGA for primary care in resource-limited settings. Methods: Longitudinal study, with median follow-up time of 16 months. Older adults from a public primary care unit in São Paulo, Brazil, were consecutively admitted. Reliability was tested in a sample from a public geriatric outpatient clinic. Participants were classified by the AMPI-AB score as requiring a low, intermediate or high complexity of care. The Physical Frailty Phenotype was used to explore the AMPI-AB's concurrent validity. Predictive validity was assessed with mortality, worsening of the functional status, hospitalizations, emergency room (ER) visits and falls. The area under the ROC curve and logistic regression were calculated for binary outcomes, and a Cox proportional hazards model was used for survival analysis.(Continued on next page)