The average chemical composition of juvenile upper continental crust (UC) as a function of age is estimated from chemical analyses, geologic maps, stratigraphic sections and isotopic ages. High plutonic/supracrustal ratios in Archean UC may reflect in part a different mode of crustal formation prior to 2.5 Ga. Greenstones show an increase in felsic volcanics and graywackes in post-Archean successions, a drop almost to zero in the proportion of komatiites at the Archean-Proterozoic (A/P) boundary, and an increase in the proportion of andesites in the Phanerozoic. Compared to the Early Archean, Late and post-Archean UC are depleted in Mg, Cr, Ni and Co, and post-Archean UC is enriched in LILE (large-ion lithophile elements; K, Rb, Ba, Th, U), HFSE (high field strength elements; Ti, P, Zr, Hf, Nb, Ta, Y), and HREE (heavy rare-earth elements). Negative Eu anomalies characterize UC of all ages, although they are relatively small in Archean UC. Cratonic shales show the same secular compositional changes as UC with the following important exceptions: At the A/P boundary, only shales show a decrease in Fe, V and Sc and an increase in Na, Ca and Sr, and only UC shows a significant increase in HREE and Y. Although post-Archean UC has a larger Eu anomaly than Archean UC, it is clear that both Archean UC and Archean shales have negative Eu anomalies, and that shales give only a weak indication of an increasing Eu anomaly in UC formed after the Archean. In both UC and shales at the A/P boundary, Cr/Th, Ni/Co and Co/Th ratios decrease, whereas the Th/U ratio increases only in shales and not in UC. The lack of HREE depletion and the high Fe, V and Sc contents of Archean shales indicate these shales were not derived from sources similar in composition to exposed Archean UC. The shale sources, which are now removed by erosion, must have been composed largely of basalt (± komatiite) (60%) and granite (40%), with little contribution from HREE-depleted TTG (tonalite-trondhjemite-granodiorite), which now dominates in Archean exposures. This suggests strong vertical zonation in the Archean continents with granites intruded into greenstones at very shallow levels. Compositional changes in UC or/and shales at the A/P boundary can be explained in one of the four ways: (1) a greater amount of basalt and komatiite in Archean UC (high Cr and Ni in Archean UC and shales); (2) garnet/amphibole fractionation during production of TTG by partial melting of hydrous basaltic crust (low HREE and Y in Archean UC); (3) change in TTG magma source from descending slab or thickened crust to metasomatized mantle wedge (increase in LILE and LREE in post-Archean UC and shales); and (4) a decrease in the intensity of chemical weathering after the Archean (increase in Na, Ca and Sr in post-Archean shales).