In a very important article Paul Thieme demonstrated that Ved. Sindhu- was a nominal -u- formation, based on a non-attested present stem *sinadh-/*sindh- derived by a root sidh (usual present stem sadha-). Then, *sindhú- would mean ‘warding off, keeping away’, while the corresponding substantivization as síndhu-, m./f., (with the stress on the root) meant ‘he/she who wards off, keeps away’, i.e. ‘natural frontier’. Notwithstanding the evident absence of a present stem like *sinadh-/*sindh-, the reconstruction suggested by Thieme is still the most reasonable in the light of the Ṛgvedic passages where the older interpretation of síndhu- just as ‘river’ or ‘ocean’ results patently far-fetched or simply impossible. In any case, for the Avestan corresponding form, hiṇdu-/həṇdu- (O.P. hindu-), no good Iranian or Indo-Iranian (independent) etymology seems to be recognizable, and it is presumable, as normally stated, that this one was a very early loanword in Iranian, just meaning not ‘India’, but ‘(natural) frontier’, and thus referring to barriers or obstacles as big and large as a river or a large basin of water (lake, sea or ocean). Here some problems connected with this interpretation of the linguistic data are analyzed, with special regard for the Indo-Iranian and Avestan mythology.