Abstract: Abstract This paper employs phylogenetic modeling to reconstruct the alignment system of Indo-European. We use a data set of categorical morphosyntactic features, which take states such as ‘nominative-accusative’, ‘active-stative’, or ‘ergative’. We analyze these characters with a standard Bayesian comparative phylogenetic method, inferring transition rates between character states on the basis of a phylogenetic representation of the languages in the data. Using these rates, we then reconstruct the probability of presence of traits at the root and nodes of Indo-European. We find that the most probable alignment system for Proto-Indo-European is a nominative-accusative system, with low probabilities of neutral marking and ergativity in the categories lower in grammatical hierarchies (nouns, past). Using a test of phylogenetic signal, we find that characters pertaining to categories higher in hierarchies show greater phylogenetic stability than categories lower in hierarchies. We examine our results in relation to theories of Proto-Indo-European alignment as well as to general typology.