AIMS: Establishing whether fasting plasma glucose (FPG), postprandial glucose (PPG), and HbA1c have the same diagnostic accuracy in NAFLD versus otherwise healthy people could inform T2DM screening recommendations for those with NAFLD. METHODS: Cross-sectional analysis of the Third National Health and Nutrition Examination Survey (NHANES III) 1989-1994. T2DM was defined as PPG ≥ 200 mg/dL, FPG ≥ 126 mg/dL, or HbA1c ≥ 6.5 %. We estimated sensitivity and specificity between the six pairwise combinations between the three T2DM definitions in people with and without NAFLD. With Poisson regressions, we investigated if people with NAFLD were more likely to have T2DM with two diagnostic criteria yet not with the third one. RESULTS: There were 3652 people with mean age 55.6 years and 49.4 % were men; 673 (18.4 %) people had NAFLD. Compared to NAFLD-free individuals, those with NAFLD had lower specificity in all pairwise comparisons except when PPG was the reference vs HbA1c [98.28 % (95 % CI: 97.73 %-98.72 %) in people without NAFLD vs 96.15 % (95 % CI: 94.28 %-97.54 %)]. The sensitivity of FPG was slightly superior to PPG and HbA1c in people without NAFLD; for example, 64.62 % (95 % CI: 55.75 %-72.80 %) for FPG vs 56.58 % (95 % CI: 44.71 %-67.92 %) for HbA1c. People with NAFLD were more likely to be diagnosed with FPG and PPG yet not with HbA1c (PR=2.15; p = 0.020). CONCLUSIONS: While these T2DM diagnostic criteria may capture different patients both in people with and without NAFLD, in the NAFLD population FPG appears to have the best sensitivity and there were no differences between PPG and HbA1c in terms of specificity.