The increased mortality risk of hyperhomocysteinaemia in diabetes may be mitigated by dietary quality.The Nutrition and Health Survey in Taiwan of 1999-2000 for elders formed this prospective cohort. Baseline health status, diet and anthropometry were documented and plasma homocysteine and biomarkers for B vitamins measured. Participants without diabetes (n=985) were referent for those who had diabetes or developed diabetes until 2006 (n=427). The effect of homocysteine on mortality risk during 1999-2008 was evaluated.Men, smokers and those with poorer physical function had higher homocysteine, but less so with diabetes. Diabetes incidence was unrelated to homocysteine. In hyperhomocysteinaemia (>=15 vs <15 μmol/L), those with diabetes had an adjusted hazard ratio (HR) (95% CI) for mortality of 1.71 (1.18-2.46); p for interaction between homocysteine and diabetes was 0.005. Without diabetes, but with hyperhomocysteinaemia and a low dietary diversity score (DDS <=4 of 6), where the joint mortality hazard for the greater DDS, (>4) and lower homocysteine (<15) was referent, the HR was 1.80 (1.27-2.54) with significant interaction (p=0.008); by contrast, there was no joint effect with diabetes. The contribution of DDS to mortality mitigation in hyperhomocysteinaemia could not be explained by B group vitamins, even though plasma folate was low in hyperhomocysteinaemic participants. With hyperhomocysteinaemia, heart failure was a major cause of death.In non-diabetic hyperhomocysteinaemia, a more diverse diet increases survival prospects independent of B group vitamins, but not in hyperhomocysteinaemic diabetes where the cardiomyopathy may be less responsive.