Analysing correlations between the observed health effects of ingested probiotics and their survival in digestive tract allows adapting their preparations for food. Tracking ingested probiotic in faecal samples requires accurate and specific tools to quantify live vs dead cells at strain level. Traditional culture-based methods are simpler to use but they do not allow quantifying viable but non-cultivable (VBNC) cells and they are poorly discriminant below the species level. We have set up a viable PCR (vPCR) assay combining propidium monoazide (PMA) treatment and either real time quantitative PCR (qPCR) or droplet digital PCR (ddPCR) to quantify a Lactobacillus rhamnosus and two Lactobacillus paracasei subsp. paracasei strains in piglet faeces. Adjustments of the PMA treatment conditions and reduction of the faecal sample size were necessary to obtain accurate discrimination between dead and live cells. The study also revealed differences of PMA efficiency among the two L. paracasei strains. Both PCR methods were able to specifically quantify each strain and provided comparable total bacterial counts. However, quantification of lower numbers of viable cells was best achieved with ddPCR, which was characterized by a reduced lower limit of quantification (improvement of up to 1.76 log10 compared to qPCR). All three strains were able to survive in the piglets’ gut with viability losses between 0.78 and 1.59 log10/g faeces. This study shows the applicability of PMA-ddPCR to specific quantification of small numbers of viable bacterial cells in the presence of an important background of unwanted microorganisms, and without the need to set up standard curves. It also illustrates the need to adapt PMA protocols according to the final matrix and target strain, even for closely related strains. The PMA-ddPCR approach provides a new tool to quantify bacterial survival in faecal samples from a preclinical and clinical trial.