Relations within the Iranian branch of Indo-European have traditionally been modelled by a tree that is essentially composed of binary splits into sub-and sub-subbranches. The present article will argue that this tree is rendered outdated by new data that have come to light from contemporary and ancient languages, and that it was methodologically problematic from the outset, both for reasons of the isoglosses on which it is based and for not taking into account distinctions such as shared innovations vs. shared archaisms and marked vs. unmarked language changes. Conversely, a set of non-trivial morphological innovations, particularly in the verbal system, shared by Bactrian, Parthian and some neighbouring languages appears to suggest a subbranch which I will call "Central Iranian". The second part of the paper will present a tentative reconstruction of the nominal system of Central Iranian. This reconstruction wishes to show the result one arrives at when trying to reconstruct a subbranch as strictly bottom-up as possible, i.e. using only the data from the languages under study, and avoiding to profit from Old Iranian data and from our knowledge about the protolanguages. " Central Iranian " would be closely related to Sogdian, and would share, by language contact, a number of features with Middle Persian; at the same time, Bactrian and Sogdian also share features with individual Eastern Ir. languages.