Background: There has been uncertainty about the risk of new-onset diabetes in hypertensive individuals treated with different blood pressure-decreasing drugs. Objectives: To study this risk in hypertensive individuals who were at risk of developing diabetes mellitus in the Losartan Intervention For Endpoint reduction in hypertension (LIFE) study. Methods: In the LIFE study, with a double-masked, randomized, parallel-group design, 9193 patients (46% men) with hypertension (mean age 67 years, average pressure 174/98 mmHg after placebo run-in) and electrocardiogram-documented left ventricular hypertrophy were randomly assigned to once-daily losartan- or atenolol-based antihypertensive treatment and followed for at least 4 years (mean 4.8 years). At baseline, 7998 patients did not have diabetes mellitus and were thus at risk of developing this condition during the study. To demonstrate ability to predict new-onset diabetes, we developed a prediction score using the significant variables from multivariate analyses (serum glucose, body mass index, serum high-density lipoprotein cholesterol, systolic blood pressure and history of prior use of antihypertensive drugs). Results: There was a steadily increasing risk of diabetes with increasing level-of-risk score; patients in the highest quartile were at considerably greater risk than those in the three lower ones. Treatment with losartan was associated with lower risk of development of diabetes within each of the four quartiles of the risk score. As previously reported, new-onset diabetes mellitus occurred in 242 patients receiving losartan (13.0 per 1000 person-years) and 320 receiving atenolol (17.5 per 1000 person-years); relative risk 0.75 (95% confidence interval 0.63 to 0.88; P < 0.001). Conclusions: New-onset diabetes could be strongly predicted by a newly developed risk score using baseline serum glucose concentration (non-fasting), body mass index, serum high-density lipoprotein cholesterol concentration, systolic blood pressure and history of prior use of antihypertensive drugs. Independently of these risk factors, fewer hypertensive patients with left ventricular hypertrophy developed diabetes mellitus if they were treated with losartan than if they were treated with atenolol. © 2002 Lippincott Williams & Wilkins.