Undercarboxylated osteocalcin (ucOC) modulates insulin secretion and sensitivity in mice, and higher ucOC is associated with lower prevalence of diabetes in men. The influence of ucOC distinct from other markers of bone turnover on incidence of cardiovascular events is unclear.Community-dwelling men aged 70-89 years resident in Perth, Western Australia.Serum total osteocalcin (TOC), N-terminal propeptide of type I collagen (P1NP), and collagen type I C-terminal cross-linked telopeptide (CTX) were measured by immunoassay, and ucOC by hydroxyapatite binding. The ratio ucOC/TOC was calculated. Hospital admissions and deaths from myocardial infarction (MI) and stroke were ascertained.There were 3384 men followed for 7.0 years, during which 293 experienced an MI, 251 stroke, and 2840 neither. In multivariate analyses, higher ratio of ucOC/TOC (expressed as %) was associated with lower incidence of MI (quartiles Q2-4, ≥ 49% versus Q1,<49%, hazard ratio 0.70, 95% confidence interval = 0.54-0.91), but not of stroke (0.99, 0.73-1.34). Higher P1NP was associated with higher incidence of MI (Q2-4, ≥ 28.2 μg/L versus Q1, <28.2 μg/L, hazard ratio 1.45, 95% confidence interval = 1.06-1.97), but not of stroke (0.94, 0.70-1.26). CTX was not associated with incident MI or stroke.A reduced proportion of undercarboxylated osteocalcin or higher P1NP are associated with increased incidence of MI. UcOC/TOC ratio and P1NP predict risk of MI but not stroke, in a manner distinct from CTX. Further studies are needed to investigate potential mechanisms by which bone turnover markers related to metabolic risk and to collagen formation could modulate cardiovascular risk.