Background: The aim of the study was to determine how much of the association between obesity, measured by body mass index (BMI), and cognition in older persons is mediated through inflammation, fasting plasma glucose, and hypertriglyceridemia. Methods: Anthropometrics, high-sensitivity C-reactive protein (CRP), fasting plasma glucose, and serum triglycerides were measured in 3,256 community-dwelling individuals aged 55-85 years residing in Newcastle, New South Wales, Australia. Audio recorded cognitive screen (ARCS) was used to assess multiple cognitive domains. Results: Mediation analyses showed very modest but significant direct mediation effects, whereby obesity was associated with better cognitive function after adjusting for potential confounders (controlled direct effect ≈ 1/500 point increase in the total ARCS score per 1.0-kg/m² increase in BMI). There were significant indirect negative mediation effects from BMI to cognition mediated through CRP, that is, increased BMI was associated with increased CRP which was associated with decreased cognition (natural indirect effect -0.20 unit; 95% confidence interval [CI] -0.39, -0.02), and through fasting plasma glucose, that is, increased BMI was associated with increased fasting plasma glucose which was associated with decreased cognition (natural indirect effect -0.12 unit; 95% CI -0.24, -0.01], but not through serum triglycerides. Conclusions: There is a weak positive association between obesity and cognitive performance in older persons, which is partially antagonized by inflammation and elevated fasting plasma glucose, but not hypertriglyceridemia. Further studies are needed to elucidate whether this is due to selection bias, or truly reflects biologically complex and counter balancing pathways involved in obesity.