Two confounding factors were selected and analyzed in blood pressure estimation using pulse arrival time (PAT) for each individual. The heart rate was used as the confounding factor for the cardiac cycle, and the duration from the maximum derivative point to the dicrotic peak (TDB) in the photoplethysmogram was used as another confounding factor representing arterial stiffness. By considering these factors with PAT in multiple regression analysis, the performance of blood pressure estimation is enhanced significantly in the diastolic phase as well as in the systolic phase. The reproducibility of this method was also validated with formerly obtained regression equations from the training set. The correlation between estimated and measured blood pressure decreased a little, but the validity was still maintained (r congruent with 0.8). This shows the value of the method in non-intrusive blood pressure estimation for individual patients and may be useful for various applications.