Accurate prediction of the critical heat flux (CHF) is one of the key tasks of PWR core design and safety assessment, for the maximal allowable heat flux in the reactor core is limited by CHF. Since CHF in rod bundle cannot be predicted analytically, up-to-date predictive approach is based on empirical correlations related to the local thermal-hydraulic conditions, geometry and power distribution. However, development of CHF correlation for PWR fuel assemblies under low pressure conditions (2–10 MPa) is constrained by limited amount of experimental data points, which builds up in statistics a typical problem of small sample amounts, but requiring simultaneously high prediction accuracy. In our previous study, stepwise regression method was applied to develop a dimensional, empirical CHF correlation for PWR under low pressure conditions, termed as the advanced low pressure CHF correlation (ALPC), which successfully solves the challenge of small sample problem. However, the ALPC correlation still uses dimensional independent variables with less physical meanings, which limits its physical interpretability. In the current study, stepwise regression method was used to develop a revised, dimensionless version of the ALPC CHF correlation. First, various dimensionless, two-phase thermal-hydraulic parameters that might influence CHF were selected as candidate independent variables. With stepwise regression, the form and coefficients of the revised CHF correlation were optimized in a dynamic manner. Compared to the current ALPC correlation, the revised version developed in this study possesses a similar simple form but a much higher prediction accuracy. Revision of the ALPC correlation demonstrates clearly the advantages of utilizing dimensionless parameters as independent variables in CHF correlation, which points out a new direction of developing rod-bundle CHF correlations for engineering purpose.