Cuff-less blood pressure (BP) estimation using pulse transit time (PTT) is a promising method for long-term BP monitoring. However, state-of-art PTT models are unable to trace the change of pressure baseline in subjects, which limits their application in long-term BP tracking. This study investigated the relationship between the change of pressure baseline and pulse waveform in long-term BP monitoring. In the study, a total of 36 subjects received daily monitoring of systolic BP (SBP) and PTT for over one month. Linear regression was used to develop the SBP-ln(PTT) model for each subject. SBP predictions with regression differences greater than +/- SD (7.63 mmHg) were assumed to be with positive/negative baseline offset. For each subject, 12 features extracted from pulse waveform were obtained and their values were converted to standard scores to quantify pulse waveform variation. Independent two-sample t-test showed five pulse wave features changed significantly when subjects' pressure baseline varied. Furthermore, the consistency of pulse waveform variation was validated over the change of pressure baseline in subjects. In summary, this study demonstrated that pulse waveform could indicate baseline offset in PTT-based BP estimation. By highlighting five pulse wave features, this study provides novel insights to overcome the challenge of frequent calibrations in long-term PTT-based BP monitoring.