ObjectiveFrailty and heart failure (HF) are cross-sectionally associated. Published longitudinal data are very limited. We sought to investigate associations between frailty and incident HF.MethodsProspective study of 1722 men, examined at age 72–91 years. Scores based on the Fried phenotype, Gill index and a novel frailty score, based on the Health Ageing and Body Composition Battery, incorporating slow walking speed, low chair-stand time and subjective difficulty with balance, were calculated. Associations between these scores and incident HF were analysed with Cox proportional hazard modelling.Results1445 men with frailty data and without prevalent HF were included. 99 developed HF (mean follow-up 6.1 years). Men scoring 3/3 on our novel frailty score had elevated risk of incident HF (HR 2.77, 95% CI 1.25 to 6.15), which persisted after adjustment for established risk factors and interleukin-6 (HR 3.14, 95% CI 1.35 to 7.31). This risk remained increased, although attenuated, after excluding HF events within 2 years of baseline (HR 2.05, 95% CI 0.61 to 6.92). The frailty phenotype showed a non-significant association with HF (age-adjusted HR 1.92, 95% CI 0.99 to 3.73), which was further attenuated after adjustment for prevalent coronary heart disease and Body mass index (HR 1.60, 95% CI 0.81 to 3.15). Gill-type scores were weakly associated with HF risk after these adjustments (HR 1.31, 95% CI 0.47 to 3.70).ConclusionIn these older men, the combination of slow walk speed, low sit-stand time and balance problems were associated with high risk of incident HF, independent of established risk factors and inflammatory markers. However, undiagnosed HF at baseline may still be a confounder. There is a differential association between aspects of the frailty phenotype and incident HF.