Background and Aims Type 2 diabetes mellitus (T2DM) is a condition defined by hyperglycaemia, but also often presents with dyslipidaemia and suppressed HDL cholesterol. Mendelian randomization studies have suggested a causal link between low HDL cholesterol and T2DM. However, influences of gender, polymorphisms and lifestyle, all known to influence HDL cholesterol, have not been fully explored in a prospective cohort. Methods and Results In 2001-2002, a random sample of 1514 males (18-87 years old) and 1528 females (18-89 years old) were recruited in the ATTICA study. The 10-year follow-up (2011-2012) included 1485 participants. Lipids and lipoproteins levels, glucose and insulin levels were measured together with apolipoprotein A1 (apoA1) 75 G/A genotype, which is known to influence HDL-cholesterol. In total, 12.9% of the study sample developed T2DM within the 10-year follow-up period. In multivariable models, for each mg/dL increase in apoA1 levels in males, 10-year T2DM risk decreased 1.02%; while every unit increase in apoB/LDL-cholesterol ratio increased risk 4-fold. Finally, for every unit increase in triglycerides/apoA1 ratio, the risk increased 85%. HOMA-IR independently predicted T2DM 10-year incidence only for carriers of GG polymorphism (all, p<0.05), but not in carriers of the GA polymorphism (all, p>0.05). Conclusion ApoA1 was associated with decreased T2DM risk and TG/ApoA1 and apoB/LDL were associated with increased risk of T2DM, only in males. ApoA1 polymorphism, which is associated with lower HDL cholesterol, influenced the predictive effects of HOMA-IR on T2DM incidence, which appeared to be moderated by physical activity, suggesting potential scope for more targeted preventative strategies.