The efficacy of different structural models for describing the observed neutron-powder-diffraction (NPD) measurements of bulk polycrystalline YD 3 as well as other hexagonal rare-earth (i.e., Nd, Tb, Dy, Ho, Er, and Tm) trideuteride powders has been investigated via Rietveld refinement. Between the two possible structural configurations, centrosymmetric P 3 3c1 and noncentrosymmetric P6 3 cm, the latter can be excluded due to very high correlations found between the positions of the D sites. Hence, the true "diffraction-average" structure for YD 3 and all other rare-earth deuterides studied is centrosymmetric (P 3 3c1). This seems to contrast with the prior evidence from first-principles calculations and various spectroscopic probes suggesting that the true local symmetry is not P 3 3c1, but rather, noncentrosymmetric. A possible way to reconcile the apparently conflicting conclusions from NPD and spectroscopic measurements is by assuming that the real structure is a twinned arrangement of nanosized, noncentrosymmetric configurations. For example, we demonstrate that the diffraction-average centrosymmetric P 3 3c1 structure can result from a superposition of individual, noncentrosymmetric P3c1 twins. A comparison of neutron vibrational spectra for YH 3 and YD 3 confirms that both compounds share similar structural arrangements.