Glucose perturbations can be detected by fasting plasma glucose (FPG), HbA1c, and the oral glucose tolerance test (OGTT). The highest yield is provided by OGTT. HbA1c is considered more practical. We compare the diagnostic and predictive performance of these glycaemic indicators based on combined data from the EUROASPIRE IV (EAIV) and V (EAV) studies. This cohort study was conducted in 79 centres in 24 European countries (EAIV) and 131 centres in 27 European countries (EAV). Eligible patients were aged 18-80 years, did not have diabetes, and were diagnosed with coronary artery disease 6-36 months (EAIV) or 6-24 months (EAV) before the investigation. Patients were investigated with OGTT (FPG and 2 h post-load glucose [2-hPG]) and HbA1c. Follow-up of subsequent cardiovascular events was done by means of a questionnaire at least 1 year after the baseline investigation. Analyses were done in patients with both OGTT and HbA1c data available. Outcome analysis in these patients was restricted to those with valid follow-up data available. 16 259 patients were interviewed in EAIV (2012-13) and EAV (2016-17). 8364 patients had both OGTT and HbA1C data and were included in the analysis population (3932 in EAIV and 4432 in EAV). Information on cardiovascular events was available in 7892 patients. Follow-up was for a median 1·6 years (IQR 1·2-2·0). The average patient age was 63·3 years (SD 9·8), and 6346 (75·9%) of 8364 patients were men. At baseline, 1856 (22·5%) of 8263 patients were determined to have newly detected type 2 diabetes using OGTT alone, compared with 346 (4·2%) using HbA1c alone. New dysglycaemia, defined as newly detected type 2 diabetes or impaired glucose tolerance (IGT), was present in 3896 (47·1%) of the patients according to 2hPG. 2hPG 9 mmol/L or greater (162 mg/dL, adjusted hazard ratio [aHR] 1·58; 95% CI 1·27-1·95, p<0·0001), and HbA1c 5·9% or greater (41 mmol/mol, aHR 1·48, 1·19-1·84; p=0·0010) were the strongest predictors of cardiovascular events, while FPG did not predict. A multivariable model showed that the effect of HbA1c on cardiovascular events was mainly explained by 2hPG (aHR for 1 unit increase in HbA1c 1·13, 0·98-1·30; p=0·11; and aHR for 1 unit increase in Ln[2hPG] 1·37, 1·08-1·74; p=0·0042). 2hPG appears better than HbA1c in detecting dysglycaemia and predicting its impact on future cardiovascular events in patients with coronary artery disease and should be recommended as the primary screening tool. Swedish Heart-Lung Foundation, Region Stockholm (ALF), the Erling Persson Foundation, the Baltic Child Foundation.