It is unclear which blood pressure (BP) components (that is, systolic BP (SBP), diastolic BP (DBP), pulse pressure (PP) and mean arterial pressure (MAP)) are superior predictors of chronic kidney disease (CKD). Furthermore it is unclear whether the combination of SBP+DBP or PP+MAP is superior to any of these four individual BP components in predicting CKD. We enrolled 9928 Japanese men aged 40-55 years who had a normal estimated glomerular filtration rate (eGFR), no proteinuria and no history of cardiovascular disease and were not taking any antihypertensive medications at baseline. CKD was defined as an eGFR of o60 ml min À1 per 1.73 m 2 using the modified diet in renal disease equation. DAkaike's information criterion (DAIC) was used to compare the BP components-added model to the model without them in a Cox proportional hazards model. During the 52 428 person-years of follow-up, there were 434 cases of CKD. Of all four BP components, the model including DBP-or MAP-alone had the highest values of DAIC (10.2 and 9.85, respectively). The PP-alone model had the lowest DAIC value (À1.48). The combination models including SBP+DBP (DAIC 8.42) or PP+MAP (8.42) were not superior to the models including DBP-or MAP-alone. These findings suggested that, of the four BP components, both DBP and MAP were the most useful predictors for subsequent incidence of CKD, but PP was not an important predictor. The combination model, including SBP+DBP or PP+MAP, was not superior to the models including DBP-or MAP-alone for predicting CKD.