Background Biomarkers may contribute to risk stratification in coronary heart disease (CHD). We examined whether plasma midregional proadrenomedullin (MR-proADM) concentration at baseline and its change over one year predicts long-term outcomes in stable CHD patients. Methods The LIPID study randomised patients 3-36 months after an acute coronary syndrome with total cholesterol 4.0-7.0 mmol/L (155-271 mg/dL), to placebo or pravastatin 40 mg. Follow-up was 6.0 years. MR-proADM plasma concentrations at baseline and one year later were determined in 7863 and 6658 patients, respectively. These were categorised into quartiles to perform Cox regression analysis, adjusting for baseline parameters. Results Baseline MR-proADM concentrations predicted major CHD events (non-fatal myocardial infarction or CHD death; hazard ratio (HR) 1.52, 1.26-1.84 for Q4-Q1), CHD death (HR 2.21, 1.67-2.92), heart failure (HR 2.30, 1.78-2.97) and all-cause mortality (HR 1.82, 1.49-2.23). Associations were still significant after adjustment for baseline B-type natriuretic peptide (BNP) concentration. Increase in MR-proADM after one year was associated with increased risk of subsequent CHD events (HR 1.34, 1.08-1.66), non-fatal myocardial infarction (HR 1.50, 1.12-2.03), heart failure (HR 1.78, 1.37-2.30) and all-cause mortality (HR 1.31, 1.04-1.64). Associations with heart failure and all-cause mortality remained significant after adjusting for baseline and change in BNP concentration. Change in MR-proADM moderately improved risk reclassification for major CHD events (net reclassification improvement (NRI) 3.48%) but strongly improved risk reclassification for heart failure (NRI 5.60%). Conclusions Baseline and change in MR-proADM concentrations over one year are associated with risk of major clinical events, even after adjustment for BNP concentrations.