Magmatic activity linked to syn- or post-collisional zones leads to the emplacement of remarkably heterogeneous rocks: calc-alkaline, high-K calc-alkaline or shoshonitic series variably contaminated by continental crust; anatectic granites and ignimbrites derived from the latter; and finally alkali potassic to ultrapotassic basalts [Harris et al., 1990; Pearce et al., 1984, 1990; Arnaud et al., 1992; Benito et al., 1999]. The main sources of these magmas are either the upper mantle (sub-oceanic or subcontinental) frequently metasomatized by hydrous fluid originating from the subducted slab; or the continental crust, which can act as a contaminant [Benito et al., 1999; Miller et al., 1999] or melt directly [Harris et al., 1990; Zingg et al., 1990]. The purpose of the present paper is to document the role of a third source: the subducted oceanic crust, as evidenced by the occurrence of Miocene adakites in Sarawak (NW Borneo). The studied rocks have been sampled from western Sarawak (fig. 1), and their location is shown on the geological map [Tan, 1982] of figure 2. They mostly occur as stocks, dykes and sills which crosscut the Paleozoic to Miocene sedimentary units. Two kinds of intrusions can be distinguished. High-K calc-alkaline to medium-K calc-alkaline diorites and microdiorites occur in the northern part of the studied area, in Salak Island and Santubong Peninsula. Microtonalites and dacites occur near Kuching and in the southern part of Sarawak (Kuap and Bau areas). Whole-rock K-Ar data (table I) demonstrate that these two associations are of different ages: high-K calc-alkaline diorites were emplaced during the Lower Miocene (22.3 to 23.7 Ma), whereas the microtonalites and dacites are younger by ca. 8 Ma or more (Middle to Upper Miocene, 14.6 to 6.4 Ma). Major and trace element data (table II) show that the Lower Miocene diorites display all the usual characteristics of subduction-related magmas. The Middle to Upper Miocene microtonalites and dacites share some of these characteristics, but in addition they display typical adakitic features: SiO 2 -rich (65.5-70%) and sodic (Na 2 O/K 2 O>2) character (table II and figure 3); lack or rare occurrence of pyroxenes, usually replaced by early-crystallized (near-liquidus) amphiboles (table III); very low Y and HREE contents, consistent with the presence of residual garnet in their source, and leading to characteristically high La/Yb and Sr/Y ratios (fig. 4, 5). Their titanomagnetite-hemoilmenite associations reflect equilibrium features [Bacon and Hirschman, 1988] indicating moderate temperatures (<900 degrees C) and highly oxidizing (NNO+1) crystallization conditions [Ghiorso and Sack, 1991]. The Lower Miocene Sarawak diorites are typically subduction-related from a geochemical point of view. They likely derive from the evolution of island-arc basaltic magmas, which themselves originated from the partial melting of upper mantle peridotites previously metasomatized by hydrous fluids expelled from the subducting oceanic slab [Tatsumi et al., 1986; Tatsumi, 1989]. The origin of the Middle-Upper Miocene adakitic microtonalites and dacites is different. According to previous studies, they likely derive from the partial melting of metabasalts (garnet amphibolites or eclogites) from subducted oceanic crust [Defant and Drummond, 1990; Defant et al., 1991, 1992; Drummond et al., 1996; Maury et al., 1996; Martin, 1993, 1999]. Their position in the hybrid tonalite+peridotite system [Caroll and Wyllie, 1989] shows that they crystallized within the garnet stability field and likely interacted with the upper mantle during their ascent (fig. 7). This feature is not consistent with their genesis through melting of metabasalts accreted at the base of the Borneo continental crust. In addition, the less evolved Sarawak adakites display mineralogical and geochemical features remarkably similar to those of the 1991 Mt Pinatubo dacite, the experimental petrology of which has been extensively studied at low [2 kbar; Scaillet and Evans, 1999; Rutherford and Devine, 1996] to medium pressures [4 to 20 kbar; Prouteau et al., 1999]. Such dacitic magmas are not in equilibrium with garnet at pressures lower than or equal to 20 kbar, which rules out their derivation from metabasalts tectonically or magmatically accreted to the base of the North Borneo continental crust. We propose, instead, that they originated from the partial melting of basalts from a fragment of oceanic lithosphere within the upper mantle. Like the adakites of Central Mindanao, Philippines [Sajona et al., 1994, 1997 and 2000; Maury et al., 1996] and those from Aird Hills, Papua-New Guinea [Smith et al., 1979; Defant and Drummond, 1990] the Sarawak adakites represent potential markers of the occurrence at depth of oceanic crust slivers, which could be much more common in collision zones than previously thought.