AIMS: To evaluate, in hypertensive patients, whether the metabolic syndrome is a better predictor of new-onset diabetes compared with impaired fasting glucose, obesity or its other individual components alone, or collectively. METHODS: Cox models were developed to assess the risk of new-onset diabetes associated with the metabolic syndrome after adjusting for a priori confounders (age, sex, ethnicity and concomitant use of non-cardiovascular medications), its individual components and other determinants of new-onset diabetes. Area under receiver operator curves using the metabolic syndrome or models of impaired fasting glucose were compared, and the ability of these models to correctly identify those who (after 5-years of follow-up) would or would not develop diabetes was assessed. RESULTS: The metabolic syndrome adjusted for a priori confounders and its individual components, and further adjusted for other determinants, was associated with significantly increased risk of new-onset diabetes [1.19 (1.00-1.40), P = 0.05 and 1.22 (1.03-1.44), P = 0.02, respectively]. The discriminative ability of the metabolic syndrome model [area under receiver operating curve: 0.764 (0.750-0.778)] was significantly better than the model of impaired fasting glucose [0.742 (0.727-0.757)] (P < 0.001). The metabolic syndrome correctly allocates the risk of new-onset diabetes in a significantly higher proportion of patients (62.3%) than impaired fasting glucose status (37.7%) (P < 0.001). The presence of both the metabolic syndrome and impaired fasting glucose were associated with an approximately 9-fold (7.47-10.45) increased risk of new-onset diabetes. Among normoglycaemic patients, the metabolic syndrome was also associated with significantly increased risk of new-onset diabetes, after adjusting for BMI and a priori confounders [1.66 (1.29-2.13)]. CONCLUSIONS: Both impaired fasting glucose and the metabolic syndrome predict the risk of new-onset diabetes; however, the metabolic syndrome is a better predictor than impaired fasting glucose in assigning the risk of new-onset diabetes in hypertensive patients, and among those with normoglycaemia.