In order to decipher the causes of Sm-Nd isotopic disequilibrium in high-pressure, low-temperature rocks, Sm-Nd isotopic analyses were carried out on minerals from four Himalayan (Tso Morari unit) and four Alpine (Dora-Maira, Monte Viso, Sesia Lanzo) eclogitic rocks of different lithologies and different intensities of post-eclogitic metamorphism. In most of these samples, garnets show the following striking features: either 1 4 7 Sm/ 1 4 4 Nd ratios lower than whole-rock, or 1 4 3 Nd/ 1 4 4 Nd ratios lower than whole-rock for higher 1 4 7 Sm/ 1 4 4 Nd ratios. In both cases, no age can be calculated. Two hypotheses can be proposed to explain the strong isotopic disequilibrium in garnet separates: (1) the preservation of the isotopic signature of the plagioclase from which garnet formed, (2) the occurrence in garnet of inherited sub-microscopic crustal inclusions with low 1 4 7 Sm/ 1 4 4 Nd and 1 4 3 Nd/ 1 4 4 Nd ratios. Knowing the Lu-Hf eclogitisation age for two samples, these two hypotheses are simulated: (1) the composition of plagioclase before eclogitization is recalculated from the measured garnet composition. The plagioclase/whole-rock age obtained with this simulation is geologically meaningless, showing that this hypothesis is invalid, (2) the 1 4 3 Nd/ 1 4 4 Nd ratio of sub-microscopic inclusions is calculated considering that the garnet separate, which was analysed, is a mixture between inclusions with low 1 4 7 Sm/ 1 4 4 Nd ratios (0.05-0.1) and pure garnet with a 1 4 7 Sm/ 1 4 4 Nd of 2. The inclusion/whole-rock ages recalculated for both samples suggest that inclusions have an old, crustal origin. The hypothesis of sub-microscopic inclusions in garnet can therefore explain the observed isotopic disequilibrium provided that the inclusions have a low Sm/Nd ratio, like allanite, apatite (and monazite), and that they are inherited from crustal contamination processes. In the case of a magmatic protolith, contamination can be achieved through crustal assimilation or a first stage of circulation of crustal fluid. In the case of Dora-Maira, the possible sedimentary protolith contains components from an old contaminant crust.In the Himalayan samples, the inverse correlation between the degree of retrogression in blueschist and then amphibolite facies, and the extent of disequilibrium between garnet and other mineral phases suggest a second stage of fluid circulation associated with retrograde metamorphism. This fluid circulation would result in the dissolution of these sub-microscopic crustal inclusions and partial leaching of their rare-earth elements (REEs) from the rock, which would scavenge the crustal 1 4 3 Nd/ 1 4 4 Nd isotopic signature from the rock.