The biogeography, chromosome number evolution, pollination biology and evolutionary history of the plant family Araceae have recently become much clearer (Cabrera et al., 2008; Chartier et al., 2013; Cusimano et al., 2011, 2012; Nauheimer et al., 2012). However, phylogenetic ambiguity near the root of the tree precludes answering questions about the early evolution of the family. We use Illumina sequencing technology and reference based assembly to resolve the remaining questions in the deep phylogeny of Araceae. We sampled 32 genera and obtained 7 from GenBank (including an outgroup), representing 42 of 44 major clades described in Cusimano et al. (2011). A subsequent phylogenomic analysis based on mitochondrial data was performed to test congruence between plastid and mitochondrial data for phylogenetic inference. Plastid sequences produced strongly supported phylogenies. In contrast, mitochondrial phylogenies were weakly supported and incongruent with chloroplast data (Templeton test, p⩽0.0001), although several smaller clades were recovered. New strongly-supported clades seen here are: (1) Anubias and Montrichardia, excluding Calla, form a clade that is sister to the Zantedeschia clade; (2) the South African genus Zantedeschia is sister to the Old World Anchomanes clade; and (3) within the Zantedeschia clade, Philodendron is sister to the rest. Calla and Schismatoglottis form a clade at the base of one of two major clades in Aroideae based on complete chloroplast sequences. Although statistical support is weak, morphological and cytological features support this topology.