Timescales of magma transfer and differentiation processes can be estimated when the magma differentiation mechanism is known. When conventional major-and trace-element analyses fail to distinguish between various processes of magma differentiation, isotope compositions can be useful. Lower Th isotope ratios in silicic relative to basaltic magmas at a given volcano, could either result from magma storage over a period of several tens of thousands of years of if the differentiation process was fractional crystallisation alone, or from crustal anatexis on a much shorter timescale. Recently mapped bimodal tephra layers from Mt. Hekla, Iceland, confirm lower ( 230 Th/ 232 Th) and higher Th/U in silicic versus mafic magmas. Higher Th/U has been taken to indicate either apatite fractionation or partial crustal melting. In-situ trace element analysis of apatite and the enveloping glass in basaltic andesite, dacite and rhyolite was undertaken to examine its capacity to fractionate trace elements and their ratios. Both Th and U are compatible in apatite with a partition coefficient ratio D U /D Th of 1. Hence, apatite crystallization and separation from the melt has a negligible effect on Th/U in Hekla magmas. Partial melting of hydrothermally altered crust remains the preferred mechanism for producing silicic melt beneath Hekla. Ten to twenty percent partial melting of metabasaltic crust with 0.4-1.2 wt% H2O produce dacite magma with 4-6% water. Absence of low δ 18 O values in Hekla magmas compared to silicic magmas of the rift-zones suggests mild hydration of the hydrothermally altered crust. Silicic magma formation, storage, differentiation and eruption at Hekla occurred over a timescale of less than a few centuries.Decreasing production of rhyolite and dacite during the Holocene lifetime of Hekla suggests changes in the crustal magma source and readjustment of the magma system with time.