Abstract Neotropical Maranta has repeatedly emerged as non-monophyletic in molecular phylogenetic studies, but no taxonomic changes have been proposed due to previous weak support for the main clades and overall sparse taxonomic sampling. Our study includes a phylogenetic hypothesis strictly based on molecular evidence, using nuclear ribosomal (nrITS) and plastid (rps16, trnL-F and rpl32-trnL) markers for Maranta and allied genera. Thirty-two species from eight genera were sampled, and maximum likelihood (ML) and Bayesian inference (BI) analyses were carried out. Non-ambiguous indels were scored in both analyses to test their contribution to internal support. Our results confirm that Maranta, as previously delimited, is non-monophyletic, with species of Hylaeanthe, Myrosma and Koernickanthe nested among clades of Maranta. The combined BI analysis without indels was the best resolved, and inclusion of indels increased support only for terminal clades. The sampled species comprise two sister clades, one with Maranta + Hylaeanthe + Myrosma + Koernickanthe and the other with Ctenanthe + Saranthe + Stromanthe. The infrageneric classification proposed by Schumann for Maranta (M. subgenera Maranta and Calateastrum) is partially corroborated by our results, but the remaining subgenera need to be further studied. Based on our strongly supported phylogenetic results, we revise the taxonomy of these genera, expanding the limits of Maranta. We propose two new combinations and a new name for Hylaeanthe in Maranta.