Background: In small, short-term studies, acute administration of caffeine decreases insulin sensitivity and impairs glucose tolerance.

Objective: To examine the long-term relationship between consumption of coffee and other caffeinated beverages and incidence of type 2 diabetes mellitus.

Design: Prospective cohort study.

Setting: The Nurses' Health Study and Health Professionals' Follow-up Study.

Participants: The authors followed 41 934 men from 1986 to 1998 and 84 276 women from 1980 to 1998. These participants did not have diabetes, cancer, or cardiovascular disease at baseline.

Measurements: Coffee consumption was assessed every 2 to 4 years through validated questionnaires.

Results: The authors documented 1333 new cases of type 2 diabetes in men and 4085 new cases in women. The authors found an inverse association between coffee intake and type 2 diabetes after adjustment for age, body mass index, and other risk factors. The multivariate relative risks for diabetes according to regular coffee consumption categories (0, <1, 1 to 3, 4 to 5, or > or =6 cups per day) in men were 1.00, 0.98, 0.93, 0.71, and 0.46 (95% CI, 0.26 to 0.82; P = 0.007 for trend), respectively. The corresponding multivariate relative risks in women were 1.00, 1.16, 0.99, 0.70, and 0.71 (CI, 0.56 to 0.89; P < 0.001 for trend), respectively. For decaffeinated coffee, the multivariate relative risks comparing persons who drank 4 cups or more per day with nondrinkers were 0.74 (CI, 0.48 to 1.12) for men and 0.85 (CI, 0.61 to 1.17) for women. Total caffeine intake from coffee and other sources was associated with a statistically significantly lower risk for diabetes in both men and women.

Conclusions: These data suggest that long-term coffee consumption is associated with a statistically significantly lower risk for type 2 diabetes.