This prospective study evaluated whether baseline cholesterol efflux is associated with future development of type 2 diabetes (T2DM) in cardiovascular patients. We measured cholesterol efflux in all CORDIOPREV study (NCT00924937) participants free of T2DM at baseline (n = 462) and assessed its relationship with T2DM incidence during a 4.5 years of follow-up. Cholesterol efflux was quantified by incubation of cholesterol-loaded THP-1 cells with the participants' apoB-depleted plasma. Disposition index was estimated as beta-cell function indicator. During follow-up 106 individuals progressed to T2DM. The cholesterol efflux/apoA-1 ratio was inversely associated with T2DM development independently of traditional risk factors (model-1, OR: 0.647, 95%CI: 0.495-0.846), and after additional adjustment for glycaemic parameters (model-2, OR: 0.670, 95%CI: 0.511-0.878). When cumulative incidence of diabetes was analysed by quartiles of cholesterol efflux/apoA-I, incidence of T2DM was reduced by 54% in subjects who were in the higher cholesterol efflux/apoA-I quartile compared to subjects in the lowest quartile (p = 0.018 and p = 0.042 for model-1 and 2). Moreover, participants who were in the higher cholesterol efflux/apoA-I presented significantly higher disposition index (β = 0.056, SE = 0.026; p = 0.035). In conclusion, HDL-cholesterol efflux normalised to apoA-I was inversely associated with T2DM development in cardiovascular patients. This association was independent of several T2DM risk factors, and may be related to a preserved beta-cell function.