The well-known human pathogenic bacterium, Vibrio cholerae, can enter a physiologically viable but non-culturable (VBNC) state under stress conditions. The differentiation of VBNC cells and nonviable cells is essential for both disease prevention and basic research. Among all the methods for detecting viability, propidium monoazide (PMA) combined with real-time PCR is popular because of its specificity, sensitivity, and speed. However, the effect of PMA treatment is not consistent and varies among different species and conditions. In this study, with an initial cell concentration of 1×108CFU/ml, time and dose–effect relationships of different PMA treatments were evaluated via quantitative real-time PCR using live cell suspensions, dead cell suspensions and VBNC cell suspensions of V. cholerae O1 El Tor strain C6706. The results suggested that a PMA treatment of 20μM PMA for 20min was optimal under our conditions. This treatment maximized the suppression of the PCR signal from membrane-compromised dead cells but had little effect on the signal from membrane-intact live cells. In addition to the characteristics of PMA treatment itself, the initial concentration of the targeted bacteria showed a significant negative influence on the stability of PMA-PCR assay in this study. We developed a strategy that mimicked a 1×108CFU/ml cell concentration with dead bacteria of a different bacterial species, the DNA of which cannot be amplified using the real time PCR primers. With this strategy, our optimal approach successfully overcame the impact of low cell density and generated stable and reliable results for counting viable cells of V. cholerae in the VBNC state.