We examined associations of neighborhood population density with 12-year changes in diabetes risk (post-challenge plasma glucose), and potential moderation by neighborhood socio-economic status (SES) among 4,816 Australians. In lower SES neighborhoods, post-challenge plasma glucose increased by 6% in low-density, remained stable in medium-density and decreased by 3% in high-density neighborhoods. In medium SES neighborhoods, glucose remained stable in high-density, but increased by 2% and 3% in medium- and low-density neighborhoods, respectively. In higher SES neighborhoods, no significant interaction effect between time and density was observed. Densification may make protective contributions for diabetes risk in lower and medium SES neighborhoods.