Clinopyroxenes from peridotite, Cr-diopside and Al-augite websterites of the Balmuccia massif have been analyzed by ion-probe for REE, Sr, Zr, Ti, V, Y, Sc and Cr. In the peridotites, the REE pattern of clinopyroxene varies from markedly LREE depleted [(Ce/Yb)_n=0.03] to moderately LREE enriched [(Ce/Yb)_n=1.96]. Clinopyroxenes of the Cr-Diopside suite vary from moderately LREE depleted to enriched [(Ce/Yb)_n=0.53–1.36], but HREE are always depleted with respect to MREE [(Sm/Yb)_n=1.19–2.06]. The clinopyroxenes of the Al-Augite dikes have higher trace element concentrations and are relatively depeleted in LREE. Two types of peridotite to websterite transitions have been documented. One type characterizes the transitions from peridotite to Cr-diopside websterite. The depletion of peridotite increases, and cpx “fertility” decreases, towards the contact; i.e. REE from Sm-Eu to Yb and Ti, V, Y, Sc and Cr in cpx define abundance trends consistent with melt extraction. LREE, Sr, and perhaps Zr, on the contrary, increase in the contact region over a distance of 0–40 cm. The variation of the latter elements and of CaO and Al₂O₃ from peridotite to websterite is smooth, whereas it is steep for the other elements. Alumina, HREE, Y and Sc are lower, Ca is higher, and the MREE/HREE pattern is steeper in the clinopyroxene of the websterite with respect to peridotite. In the second type of transition, as the contact is approached, there is an increase in modal cpx and decrease of the opx/cpx ratio. Cpx trace element patterns are similar in peridotite and websterite. This transition is generally at the Al-augite websterite contacts, but also was found at two Cr-diopside contacts. The wall rock depletion of the first transition type is interpreted as caused by melt extraction induced by intrusion of melts derived from deeper mantle (garnet-facies), whereas the LREE and Sr variation is interpreted in terms of diffusive redistribution of trace elements after emplacement of the dikes and after the modal clinopyroxene has been removed from the peridotite. Diffusive exchange occurred under stagnant conditions but in the presence of a partial melt or a “metamorphic fluid”, rather than under dry subsolidus conditions. There is no conclusive evidence favouring either a partial melt or an aqueous fluid as a medium for diffusive transport. However, we favour transport via a fluid phase because the presence of such a fluid is indicated by ubiquitous traces of amphibole in the peridotite. The second transition type is interpreted in terms of melt infiltration and re-equilibration of the melt with the peridotite matrix.