The interfacial adhesion between polymer and nanofiller plays an important role in affecting the properties of nanocomposites. The detailed relationship between interfacial adhesion and bulk properties, however, is unclear. In this work, we developed an atomic force microscopy (AFM)-based abrasive scanning methodology, as applied to model laminate systems, to probe the strength of interfacial adhesion relevant to poly(ethylene terephthalate) (PET)/graphene or clay nanocomposites. Graphite and mica substrates covered with ∼2 nm thick PET films were abrasively sheared by an AFM tip as a model measurement of interfacial strength between matrix PET and dispersed graphene and clay, respectively. During several abrasive raster-scan cycles, PET was shear-displaced from the scanned region. At temperatures below the PET glass transition, PET on graphite exhibited abrupt delamination (i.e., full adhesive failure), whereas PET on mica did not; rather, it exhibited a degree of cohesive failure within the shear-displaced layer. Moreover, 100-fold higher force scanning procedures were required to abrade through an ultimate "precursor" layer of PET only ∼0.2-0.5 nm thick, which must be largely disentangled from the matrix polymer. Thus, the adhesive interface of relevance to the strength of clay-filler nanocomposites is between matrix polymer and strongly bound polymer. At 90 °C, above the bulk PET glass transition temperature, the PET film exhibited cohesive failure on both graphite and mica. Our results suggest that there is little difference in the strength of the relevant interfacial adhesion in the two nanocomposites within the rubbery dynamic regime. Further, the bulk mechanical properties of melt mixed PET/graphene and PET/clay nanocomposites were evaluated by dynamic mechanical analysis. The glassy dynamic storage modulus of the PET/clay nanocomposite was higher than that of PET/graphene, correlating with the differences in interfacial adhesion probed by AFM.