The absence of an outcome-driven reference frame for self-measured pulse pressure (PP) limits its clinical applicability. In an attempt to derive an operational threshold for self-measured PP, we analyzed 6470 participants (mean age 59.3 years; 56.9% women; 22.5% on antihypertensive treatment) from 5 general population cohorts included in the International Database on HOme blood pressure in relation to Cardiovascular Outcome. During 8.3 years of follow-up (median), 294 cardiovascular deaths, 393 strokes and 336 cardiac events occurred. In 3285 younger subjects (<60 years), home PP only predicted all-cause and cardiovascular mortality (P⩽0.036), whereas in 3185 older subjects (⩾60 years) PP predicted total and cardiovascular mortality (P⩽0.0067) and all cardiovascular and coronary events (P⩽0.044). However, PP did not substantially refine risk prediction based on classical risk factors including mean blood pressure (generalized R(2) statistic ⩽0.20%). In older subjects, the adjusted hazard ratios expressing the risk in the upper decile of home PP (⩾76 mm Hg) versus the average risk in whole population were 1.41 (95% confidence interval, 1.09-1.81; P=0.0081) for all-cause mortality, 1.62 (1.11-2.35; P=0.012) for cardiovascular mortality and 1.31 (1.00-1.70; P=0.047) for all fatal and nonfatal cardiovascular end points combined. The low number of events precluded an analysis by tenths of the PP distribution in younger participants. In conclusion, a home PP of ⩾76 mm Hg predicted cardiovascular outcomes in the elderly with the exception of stroke, whereas in younger subjects no threshold could be established.