Abstract Background Serum bicarbonate is associated with mortality, heart failure (HF) and progression of renal failure in studies of healthy people and patients with chronic kidney disease, but the significance of these observations in unselected patients with diabetes in the general population is unknown. The aim of this study was to determine whether serum bicarbonate was associated with mortality and cardiovascular disease risk in type 2 diabetes. Methods Baseline serum bicarbonate was available for 1283 well-characterized community-based patients (mean ± SD age 64.1 ± 11.3 years, 48.7 % males) from the longitudinal observational Fremantle Diabetes Study followed for a mean of 12 years. Associations between serum bicarbonate and mortality, coronary heart disease (CHD) and incident HF were analysed using Cox proportional hazards regression. Results Serum bicarbonate was independently and negatively associated with incident CHD. For each 1 mmol/L increase in bicarbonate, the hazard ratio for CHD was 0.95 (95 % confidence interval 0.92–0.99) after adjustment for age as time scale, age at baseline, sex, English fluency, diabetes duration, log e (serum triglycerides), log e (urinary albumin: creatinine ratio), peripheral sensory neuropathy and peripheral arterial disease. There were no independent associations between serum bicarbonate and all-cause mortality [0.98 (0.95–1.004)] or incident HF [0.99 (0.95–1.03)]. Conclusions Serum bicarbonate was a significant independent predictor of incident CHD but not death or HF in community-based patients with type 2 diabetes. This supports intervention trials of bicarbonate replacement in type 2 patients at risk of CHD and who have a low serum bicarbonate concentration.