"Objectives: To examine the performance of the Achutha Menon Centre Diabetes Risk Score (AMCDRS) in predicting prevalent undiagnosed and total diabetes in Tamil Nadu, India and compare it with other Indian risk scores. Methods: AMCDRS (includes age, family history of diabetes and waist circumference) was calculated for 4896 participants (30 to 64 years) of a cross-sectional survey (2010-12) in urban and rural Vellore, Tamil Nadu. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and area under the receiver operating characteristic curve (AROC) were calculated for the optimum cut-off point. Results: Of 4896 participants, 274 (6.2%) had undiagnosed diabetes and 759 (15.5%) had total diabetes. AMCDRS score â‰¥ 4 identified 44.5% for further testing with sensitivity 75.1%, specificity 60.5%, PPV 9.1%, NPV 95.8% and an AROC of 0.639 (95% confidence interval [CI]: 0.608-0.670) for undiagnosed diabetes. The corresponding figures for total diabetes were 75.1%, 60.5%, 25.9%, 93.0% and 0.731 (95% CI: 0.713-0.750), respectively. The prevalence of undiagnosed and total diabetes increased significantly with increasing risk scores. The AROC of AMCDRS was significantly lower than that of the Chaturvedi risk score for undiagnosed diabetes but was not significantly different from AROCs of other Indian risk scores for total diabetes. Conclusions: AMCDRS, a simple risk score, can be used to identify undiagnosed and total diabetes in India. AMCDRS performed similarly in predicting total diabetes compared to other Indian risk scores which include relatively more difficult-to-measure variables (blood pressure, physical activity, body mass index). However, AMCDRS requires recalibration to improve its performance. Key words: chronic disease management and prevention, non-communicable diseases Key messages: 1. The AMCDRS is a simple diabetes risk score which was developed in a study in rural Kerala, India for predicting prevalent total diabetes, but has not been externally validated or validated for predicting undiagnosed diabetes. 2. The AMCDRS was shown to have a validity similar to other Indian risk scores (except the Chaturvedi score) which use more difficult to measure parameters, for total diabetes. 3. The lower performance of the AMCDRS for predicting undiagnosed diabetes points to the need for a recalibration of the score. "