Dependence of the ambulatory arterial stiffness index (AASI) on data scattering interferes with its potential clinical relevance. We assessed the correlates and all-cause mortality associations of a modified AASI (s-AASI). AASI was derived from the 24-h diastolic vs. systolic blood pressure linear regression line, whereas s-AASI was derived by symmetric regression (bisecting the line of diastolic vs systolic and systolic vs. diastolic). Of 2918 patients 55% were women; age was 56 +/- 16 years and body mass index was 27.3 +/- 4.5 kg/m(2). Average 24-h ambulatory blood pressure was 138 +/- 16/78 +/- 10 mm Hg. Applying the modified method for calculating AASI yielded a different measure: the negative correlation between AASI and blood pressure dipping (r = -0.304, P < 0.0001) was abolished (r = +0.223, P < 0.0001), s-AASI was more dependent on age (r = 0.266 vs. r = 0.089 for AASI), and prediction of all-cause mortality was enhanced; hazard ratio (95% confidence intervals) 1.17 (1.00-1.36) per 1 s.d. increase in s-AASI in the fully adjusted model as compared with 1.15 (0.97-1.36) for AASI.