Earthquake moment tensors and centroid locations in the catalogue of the Global CMT (gCMT) project, formerly the Harvard CMT project, have become an essential and extraordinarily valuable resource for studying active global tectonics, used by many solid-Earth researchers. The catalogue’s quality, long duration (1976–present), ease of access and global coverage of earthquakes larger than about Mw~5.5 has transformed our ability to study regional patterns of earthquake locations and focal mechanisms. It also allows researchers to easily identify earthquakes with anomalous mechanisms and depths that stand out from the global or regional patterns, some of which require us to look more closely at accepted interpretations of geodynamics, tectonics or rheology. But, as in all catalogues that are, to some extent and necessarily, produced in a semi-routine fashion, the catalogue may contain anomalies that are in fact errors. Thus, before re-assessing geodynamic, tectonic or rheological understanding on the basis of anomalous earthquake locations or mechanisms in the gCMT catalogue, it is first prudent to check those anomalies are real. The purpose of this paper is to illustrate that necessity in the eastern Himalayas and SE Tibet, where two earthquakes that would otherwise require a radical revision of current geodynamic understanding are shown, in fact, to have gCMT depths (and, in one case, also focal mechanism) that are incorrect — in spite of the overwhelming majority of gCMT solutions in that region being unremarkable and likely to be approximately correct.