Sedimentary sorting produces alternating shales and sandstones with complementary chemistries. The preferential partitioning of quartz and zircon in the coarse-grained fraction of the sediments results in a strong fractionation between SiO2, Zr, and other components (Al2O3, TiO2 ...). As a consequence, shales may be defined chemically as having higher Al2O3/SiO2 and TiO2/Zr ratios than their source rocks; sandstones have the opposite characteristics and vary in composition according to the maturity of their source and the intensity of sorting processes. Most sedimentary suites display a good correlation between Al2O3/SiO2 and TiO2/Zr. This result is used to evaluate an average bulk composition for recent clastic materials and to model their composition on a ternary Al-Ti-Zr diagram. Because the Al2O3/TiO2 ratio of igneous rocks is highly sensitive to magmatic differentiation, igneous and sedimentary trends crosscut in the Al-Ti-Zr diagram. In the case of strongly peraluminous granites (SPG), genetic models based on crustal melting and restite separation are in conflict with the observed Al-Ti-Zr relations because: (1) most SPG suites are displaced toward low TiO2/Zr ratios when compared to their potential metasedimentary sources, and (2) many SPG are too rich in Zr and Ti to be produced from calc-alkaline felsic sources in the restite model. These features suggest that the bulk Zr and Ti budget of SPG is dominated by (relatively hot) melt contributions, not by recycled solid materials.