Random phase-shifting algorithm based on principal component analysis has important application in dynamic interferometry. At present, the sign ambiguity and the reliability of the extracted phase by principal component analysis are still not be solved and valued in the exiting references. To solve these problems, we firstly extract the phase and phase shifts simultaneously from interferograms with random phase shifts by principal component analysis. Then we analyze the relationship between the extracted phase shifts and the global sign and the accuracy of the extracted phase. Finally, the sign ambiguity and the reliability of extracted phase are determined and valued by analyzing the extracted phase shifts. Numerical simulations and experiments are implemented to verify the effectiveness of this method. The simulation result shows that the global sign and the reliability of the extracted phase depend on the slope and the statistic characteristics of the phase shifts, respectively.