We determined the independent predictive value of serum apolipoprotein (apo) B/LDL-cholesterol ratio for the risk of diabetes, metabolic syndrome (MetS) and coronary heart disease (CHD).Prediction of incident cases was assessed in 2466 adults at 7 years' follow-up.ApoB/LDL ratio was independently associated log-linearly with waist circumference, and, only in men, with HDL-cholesterol in a multivariable regression model. Positive partial correlations existed with fasting insulin, fibrinogen and apo A-I and, only in women, with CRP. Cox regression analyses revealed the two highest apoB/LDL quartiles to be significant determinants of diabetes, at 2-fold RRs, independently of waist circumference, fasting glucose and other confounders. However, apoB/LDL quartiles were not independently associated with CHD in either gender. Only the highest apoB/LDL quartile was associated (RR 1.46) with the development of MetS. Serum apoB/LDL-cholesterol ratio, determined by insulin resistance and in women additionally by pro-inflammatory state, is of independent predictive value for incident diabetes and weakly for MetS, but not for CHD.