Numerous studies have shown that Mexican Americans have an increased prevalence [1][2][3][4] and incidence [5] of non-insulin-dependent diabetes mellitus (NIDDM) relative to non-Hispanic whites. The Mexican American population is characterized by increased adiposity [1][2][3][4]6], a more centralized distribution of body fat [6,7], hyperinsulinaemia [8][9][10] and insulin resistance [11,12].Metabolic factors have been shown to be important risk factors for the development of NIDDM. Both insulin resistance [13,14] and hyperinsulinaemia [13,[15][16][17][18][19] have been shown to predict NIDDM. Likewise, decreased insulin secretion in response to various stimuli may predict NIDDM. A low acute insulin response has been found to predict NIDDM in Pima Indians [13], but not in children of two Caucasian diabetic parents [14]. A low 30-min increment in insulin relative to the 30-min increment in glucose (DI 30 /DG 30 ) during an oral glucose tolerance test predicts conversion to NIDDM, especially in subjects with impaired glucose tolerance [20][21][22][23]. A low insulin response 60 min after an intravenous glucose challenge also predicted NIDDM [24,25]. A low 2-h insulin in response to an oral glucose load also Diabetologia (1997) 40: 62-66 Are risk factors for conversion to NIDDM similar in high and low risk populations? Summary Mexican Americans have an increased risk of non-insulin-dependent diabetes mellitus (NIDDM) relative to non-Hispanic whites which is only partially explained by their excess overall obesity and unfavourable body fat distribution. Non-diabetic Mexican Americans have hyperinsulinaemia and insulin resistance relative to non-Hispanic whites. We therefore hypothesized that the insulin resistance might be a more important predictor of NIDDM in high-risk populations characterized by obesity and insulin resistance, while compromised insulin secretion might be a more important risk factor for NIDDM in low-risk populations. We assessed the ability of ethnicity (Mexican American vs non-Hispanic white), age, overall adiposity (body mass index [BMI]), unfavourable body fat distribution (as assessed by waist-to-hip ratio [WHR]), glucose tolerance (impaired glucose tolerance vs normal glucose tolerance), fasting insulin and compromised insulin secretion (as assessed by increment in insulin to the increment in glucose over the first 30 min of an oral glucose tolerance test (DI 30 /DG 30 )) to predict future NIDDM. In the 8-year follow-up of the San Antonio Heart Study, NIDDM developed in 11.7 % (107/914) of Mexican Americans and in 5.0 % (18/362) of nonHispanic whites (p < 0.001). Multivariate predictors of NIDDM by multiple logistic regression analysis included increased age, BMI, WHR, fasting insulin and impaired glucose tolerance and decreased insulin secretion. The strongest independent predictors of NIDDM were high fasting insulin and decreased insulin secretion. These risk factors predicted NIDDM equally well in high and low-risk populations. [Diabetologia (1997) 40: 62-66]