In recent years, the exploration and development of shale oil and gas have expanded significantly, accompanied by numerous challenges. Porosity is a fundamental parameter for shale reservoir evaluation and reserve estimation, providing essential data for determining sweet spots and formulating development plans. Low-field nuclear magnetic resonance (LF-NMR) technology has unique non-invasive, non-destructive characteristics and has become a crucial method for shale porosity measurement. However, shale has substantially different nuclear magnetic resonance (NMR) response mechanisms from that of conventional sandstone and carbonate reservoir rocks. Furthermore, improper experimental parameters or inversion process can lead to errors or even faults in the measurement of shale porosity by NMR. In this paper, T₁-T₂ correlation is adopted to qualitatively identify hydrogen-containing components in dry and saturated shale. Based on the principle that NMR signal is proportional to the number of spin protons under uniform static magnetic field and constant temperature field, an experimental method is proposed to directly measure the shale porosity. This method calibrates with standard water sample, then compares the first amplitude of the NMR free induction decay (FID) signal between water-saturated and dry shale. The experimental results show that the NMR porosity obtained through this method is in good agreement with the weight porosity, and the influence of background signal from hydrogen-containing matrix components in shale is eliminated.