Abstract Background This study aims to investigate the predictive value of biological and neuroimaging markers to determine incident frailty among older people for a period of 5 years. Methods We included 1394 adults aged 70 years and older from the Multidomain Alzheimer Preventive Trial, who were not frail at baseline (according to Fried’s criteria) and who had at least 1 post-baseline measurement of frailty. Participants who progressed to frailty during the 5-year follow-up were categorized as “incident frailty” and those who remained non-frail were categorized as “without frailty.” The differences of baseline biochemical factors (25-hydroxyvitamin D, homocysteine, omega-3 index, C-reactive protein), other biological markers (Apolipoprotein E genotypes, amyloid-β deposits), and neuroimaging data (gray matter volume, hippocampal volume, white matter hyperintensities) were compared between groups. Cox proportional hazard model was used to evaluate the associations between biomarkers and incident frailty. Results A total of 195 participants (14.0%) became frail over 5 years. Although 25-hydroxyvitamin D deficiency, homocysteine levels, low-grade inflammation (persistently increased C-reactive protein 3–10 mg/L), gray matter, and hippocampal volume were significantly associated with incident frailty in unadjusted models, these associations disappeared after adjustment for age, sex, and other confounders. Omega-3 index was the sole marker that presented a trend of association with incident frailty (hazard ratio: 0.92; 95% confidence interval: 0.83–1.01; p = .082). Conclusions This study failed to identify biomarkers able to predict frailty incidence in community-dwelling older adults for a period of 5 years. Further longitudinal research with multiple measurements of biomarkers and frailty is needed to evaluate the long-term relationships between changes in biomarkers levels and frailty evolution.