An accurate and convenient noninvasive continue measurement of blood pressure (BP) is of great importance for the evaluation of circulatory function and prognosis of some cardiovascular diseases in out-of-hospital setting. Pulse transit time (PTT) is the most popular indicator for cuff-less BP measurement. A considerable amount of researches has demonstrated the high correlation between PTT and systolic BP (SBP) and diastolic BP (DBP). However, the greatest challenge to implement it in practice is the calibration method to get the stable and accurate correlation between BP and PTT. In this study, a new normalized PTT-BP calibration method was proposed. This method first constructed a modified calibration model exploiting nPTT (PTT normalized by RR interval) instead of PTT, and then machine learning technology was utilized to train the model. Data from 42 volunteers was collected to evaluate the proposed method. In the correlation plot, the typical correlation coefficients for SBP, DBP with the reference BP were 0.961 and 0.867 respectively. When contrast with the original model using PTT, the proposed model realized a more accurate and stable measurement of SBP, DBP. These results indicated that the presented nPTT based method could better map the dynamic relation between PTT and BP.