Reliable accounting for the diversity of microplastics (MPs) in hazard assessments involves the use of ecologically relevant metric (ERM) alignments requiring distribution data of physical particle attributes and organism bio-accessibilities. While eco-toxicological thresholds have been developed using this approach, uncertainties associated with the alignments have not been fully quantified. The present study utilized a probabilistic species sensitivity distribution (pSSD) model based on the updated Toxicity of Microplastics Explorer (ToMEx 2.0) database to derive hazard thresholds for MPs, while accounting for uncertainties due to alignments. Aligned No Observed Effect -concentrations (NOECs) were used to populate pSSDs to derive thresholds in a four-tier management framework based on the food dilution and tissue translocation-mediated effects ERMs. pSSDs differ from traditional SSDs by avoiding the use of predefined distributions to model species sensitivities, instead utilizing empirical calculations with the complete data set therefore more accurately representing underlying data. The ToMEx 2.0 dataset comprises 12,878 data points from 289 studies covering species across marine (n = 101) and freshwater (n = 67) environments. The dataset was filtered using pre-defined quality criteria, resulting in a total of 398 data points from 58 studies, with 25 marine and 19 freshwater species. Thresholds for marine and freshwater environments were derived separately using previously defined four-tier-management criteria. The resulting ERM-aligned hazard thresholds ranged from 1×10-4 to 4 particles/L, and 1 to 4,000 particles/L for marine environments, and 8 to 1,000 particles/L, and 500 to 20,000 particles/L for freshwater environments for food dilution and tissue translocation-mediated effects, respectively. Threshold uncertainties based on alignment of Monte Carlo simulations were lower for freshwater than marine environments, with percent relative standard deviations ranging from 24% to 270% for marine thresholds, and 21% to 130% for freshwater. This study is the first application of fully probabilistic methods to derive ERM-based MP hazard thresholds. We further explore the variability, uncertainty, and limitations of these thresholds through quantitative global and local sensitivity analyses, as well as qualitative discussions regarding study quality and data availability.