This study compares the plate count (PC) and the Propidium Monoazide-quantitative Polymerase Chain Reaction (PMA-qPCR) methods to assess the growth of a cocktail of three serotypes of Salmonella enterica (Heidelberg, Typhimurium, and Enteritidis) in cooked, sliced, and vacuum-packaged turkey breast (STB) under isothermal storage temperatures (8 °C–20 °C), using predictive models. Standard curves were developed for PMA-qPCR, demonstrating high efficiency (101%) and sensitivity, with quantification limits ranging from 1 to 2 log10 CFU/g for all temperatures studied. Comparative analysis revealed a significant correlation (R2 = 0.99; 95% CI) between the PC and PMA-qPCR methods; however, the agreement analysis indicated a mean difference (Bias) of −0.11 log10 CFU/g (p < 0.05), suggesting underestimation by the PC method. This indicates the presence of stressed or viable but nonculturable (VBNC) cells, detectable by PMA-qPCR but not by PC. The Baranyi and Roberts model showed a good ability to describe the behavior of S. enterica cocktail in STB for PC and PMA-qPCR data under all isothermal conditions. The exponential secondary model more accurately represented the temperature dependence of the maximum specific growth rate compared to the Ratkowsky square root model, with R2 values ≥ 0.984 and RMSE values ≤ 0.011 for both methods. These results suggest that combining PMA-qPCR with predictive modeling allows for a more accurate prediction of S. enterica growth, compared to PC method. In the event of cold chain disruptions of meat products, the use of PMA-qPCR method allow the quantification of VBNC cells, that can still pose a health risk to consumers, especially in ready-to-eat products.