Liquid-gas two-phase slug flow is the most complex flow pattern encountered in pipelines, wellbores, and cooling systems in various energy systems. Accurately predicting the slug liquid holdup (HLLS) is vital for the system design and operation because it is a crucial characteristic of slug flow that is closely related to the average liquid holdup and pressure drop. Most existing HLLS models have limited applicability due to a lack of a physical basis for slug aeration mechanisms. This study proposes a new dimensionless number SP to estimate the sharpness of entering bubbles at the liquid slug front and semi-empirically correlates SP to HLLS data, leading to the development of a unified HLLS correlation. We formulate this dimensionless number and its relations using 880 experimental HLLS data points, and the validity of the unified HLLS correlation is tested using 1050 independent data points. The developed semi-empirical model closely agrees with experimental data obtained by varying the fluid properties (0.03 mPa·s ≤ µL ≤ 966 mPa·s, 0.027 N/m ≤ σ ≤ 0.07 N/m) and operating conditions, such as the pipe geometry (1 in. ≤ pipe internal diameter ≤ 3 in., –10° ≤ θ ≤ +90°).