Aims/hypothesis Apolipoprotein C-III (apoC-III) is a small proinflammatory protein that may play a key role in diabetes pathophysiology. However, prior observational studies have been limited to predominantly white populations, and the biological links between apoC-III and diabetes, particularly the role of apoC-III on specific lipoprotein particles, are not yet well understood. We therefore investigated associations of total apoC-III and apoC-III-defined lipoprotein subspecies with incident diabetes and glucose metabolism measures in a multi-ethnic cohort. Methods For the current analyses, baseline (2000)(2001)(2002) plasma total apoC-III and apolipoprotein A-I concentrations of HDL containing or lacking apoC-III were newly measured via sandwich ELISA in 4579 participants from the Multi-Ethnic Study of Atherosclerosis. Multivariable Cox regression was used to examine associations of apolipoproteins with incident diabetes until early 2012 (567 cases), and linear mixed models were used to estimate associations with longitudinally assessed continuous measures of glucose metabolism. Similar exploratory analyses of plasma apolipoprotein B concentrations of LDL and VLDL containing or lacking apoC-III were performed in a subset of participants (LDL, n = 1545; VLDL, n = 1526). Results In the overall population, elevated total apoC-III concentrations were associated with a higher rate of diabetes (top vs bottom quintile, HR 1.88; 95% CI 1.42, 2.47; p trend = 0.0002). ApoC-III-defined HDL subspecies displayed opposing associations with incidence of diabetes (p for heterogeneity = 0.02). While HDL lacking apoC-III was inversely associated with incidence of diabetes (top vs bottom quintile, HR 0.66; 95% CI 0.46, 0.93; p trend = 0.002), HDL containing apoC-III was not associated (HR 1.11; 95% CI 0.78, 1.58; p trend = 0.61). Similarly, only HDL lacking apoC-III was beneficially associated with plasma glucose (p trend = 0.003), HbA 1c (p trend = 0.04) and insulin sensitivity (p trend < 0.0001), and higher HDL containing apoC-III was associated with lower insulin sensitivity (p trend = 0.04). Neither of the apoC-III-defined LDL subspecies was associated with incident diabetes, while VLDL was more strongly associated with the incidence of diabetes when it lacked apoC-III. Further adjustment for plasma triacylglycerols as a potential intermediate attenuated the associations of total apoC-III and apoC-IIIdefined lipoprotein subspecies. No statistically significant differences were observed across racial/ethnic groups. Conclusions/interpretation Our findings in a multi-ethnic population support the involvement of apoC-III in the development of diabetes, potentially through its association with circulating triacylglycerols. The presence of apoC-III on HDL also diminished the protective association of HDL with incident diabetes. Further investigation of apoC-III and apoC-III-defined HDL subspecies may inform the development of novel diabetes treatment and prevention strategies.Electronic supplementary material The online version of t...