The relative importance of insulin resistance and abnormal insulin secretion as risk factors for the development of non-insulin-dependent diabetes mellitus (NIDDM) is still controversial. Few data are available on insulin secretion as a risk factor for the development of NIDDM, especially in subjects with normal glucose tolerance. We examined the relation of fasting insulin (as a markerof insulin resistance) and the ratio of change in insulin to change in glucose during the first 30min after glucose ingestion (ΔI₃₀/ΔG₃₀) (as a marker ofinsulin secretion) as predictors of the 7-year development of NIDDM in 714 initially nondiabetic Mexican-Americans. NIDDM developed in 99 subjects. The relative risk of NIDDM increased with higherquartiles of fasting insulin (quartile 1 [low], 1.0; quartile 2, 1.5; quartile 3, 2.0; and quartile 4 [high], 3.7; P < 0.0001) and lower ΔI₃₀/ΔG₃₀ (quartile 1 [low], 6.9; quartile 2, 1.9; quartile 3, 1.1; quartile 4 [high], 1.0; P < 0.001). Subjects with both increased fasting insulin and decreased ΔI₃₀/ΔG₃₀ had independent increases in NIDDM incidence (P < 0.001). Further, when we stratified subjects by baseline glucose tolerance, both increased fasting insulin and decreased ΔI₃₀/ΔG₃₀ significantly predicted NIDDM in subjects with both impaired and normal glucose tolerance at baseline. We conclude that both decreasedinsulin secretion (as assessed by low ΔI₃₀/ΔG₃₀) and increased insulin resistance (as assessed by fasting insulin) predict the development of NIDDM in Mexican-Americans, a group previously characterized as having hyperinsulinemia and insulin resistance. This study provides the first evidence that decreased insulin secretion predicts the development of NIDDM in subjects with normal glucose tolerance, suggesting that deficient insulin secretion and insulin resistance occur early as a precursor of NIDDM.