In the present study, we investigated the relationship between blood pressure (BP) and pulse transit time (PTT) and evaluated the accuracy of the PTT-based cuffless BP estimation on 14 normotensive subjects. Least-squares regression was used to estimate BP in the first test and a repeatability test carried out half year later. BP in the repeatability test was also estimated using the regression coefficients in the first test. The results illustrated that in the first and repeatability tests (1) arterial BP increased and PTT decreased acutely after the exercises and (2) systolic BP was highly correlated with PTT. In the repeatability test, the estimation differences from the references were 0.0 +/- 5.3 mmHg and 0.0 +/- 2.9 mmHg for systolic and diastolic BPs respectively using least-squares regression. However, the estimation differences increased to 1.4 +/- 10.2 mmHg and 2.1 +/- 7.3 mmHg for systolic and diastolic BPs, respectively when the regression coefficients in the first test were used for prediction. In summary, reasonable BP estimations were given in the first and repeatability tests but not using the regression coefficients obtained 6 months ago for some subjects.