Summary: Genome-resolved metagenomics allows the construction of draft microbial genomes from short-read shotgun metagenomics (Metagenome-Assembled Genomes, or MAGs); however, even high-quality MAGs are typically somewhat incomplete and contain a small amount of contaminant sequence, making accurate prediction of metabolic potential challenging. Here, we describe MetaPOAP, an algorithm for probabalistic assessment of the statistical likelihoods for the presence or absence of metabolic pathways in MAGs. Availability and implementation: MetaPOAP is available as Python scripts on GitHub or from the Fischer lab webpage, https://github.com/lmward/MetaPOAP. Supplementary information: Supplementary data are available at Bioinformatics online.