Background: High-throughput sequencing provides a powerful window into the structural and functional profiling of microbial communities, but it is unable to characterize only the viable portion of microbial communities at scale. There is as yet not one best solution to this problem. Previous studies have established viability assessments using propidium monoazide (PMA) treatment coupled with downstream molecular profiling (e.g. qPCR or sequencing). While these studies have met with moderate success, most of them focused on the resulting “viable” communities without systematic evaluations of the technique. Here, we present our work to rigorously benchmark “PMA-seq” (PMA treatment followed by 16S rRNA gene amplicon sequencing) for viability assessment in synthetic and realistic microbial communities. Results: PMA-seq was able to successfully reconstruct simple synthetic communities comprising viable/heat-killed Escherichia coli and Streptococcus sanguinis . However, in realistically complex communities (computer screens, computer mice, soil and human saliva) with E. coli spike-in controls, PMA-seq did not accurately quantify viability, with its performance largely affected by community properties such as initial biomass, sample types and compositional diversity. We then applied this technique to environmental swabs from the Boston subway system. Several taxa differed significantly after PMA treatment, while not all microorganisms responded consistently. To elucidate the “PMA-responsive” microbes, we compared our results with previous PMA-based studies and found that PMA-responsiveness varied widely when microbes were sourced from different ecosystems but were reproducible within similar environments across studies. Conclusions: This study provides a comprehensive evaluation of PMA-seq exploring its quantitative accuracy in synthetic and complex microbial communities, where the technique was effective for semi-quantitative purposes in simple synthetic communities, but provided only qualitative assessments in realistically complex community samples.