Large 226Ra excesses in a number of mafic arc magmas, and geophysical observations of earthquake hypocenters locations, indicate that mafic melts can get transferred from source to surface within days to ∼1 kyr. Decreasing 226Ra-230Th disequilibria with increasing SiO2 in co-magmatic suites of individual arc volcanoes suggest that magma differentiation often occurs on timescales of a few thousand years in closed systems, or less in open systems. However, the rapid decrease of global 238U- 230Th disequilibria with increasing SiO2 within basaltic andesite compositions suggests that rapid closed system differentiation does not typically produce more evolved compositions from mafic parental magmas with large 238U excesses. In the case of rapid magmatic evolution, open system processes are frequently involved in the production of andesites and dacites, and the data imply that the open system component is close to 238U- 230Th secular equilibrium. MELTS modeling of magma evolution at Santorini in the Aegean arc provides corroborating evidence from an individual well-studied arc volcano, that closed system fractional crystallization fails to produce the more evolved compositions, and increasing 87Sr/86Sr ratios with increasing SiO2 from andesites to rhyolites suggests assimilation of old crust is involved. These observations can be conceptualized through thermal modeling of basaltic sill injections into the lower and upper crust. Melt production in deep crustal hot zones provides an explanation for low 238U-230Th disequilibria of the evolved magmatic component, as significant accumulation of evolved melts through partial re-melting of previously intruded basalts requires >100 kyr incubation time at typical magmatic flux rates. Thermal modeling also provides a framework for understanding assimilation of old upper crustal rocks in more mature magmatic systems. Further insights into magmatic evolution can be gained from dating of minerals from mafic, intermediate and felsic arc magmas. (a) Many U-Th mineral isochrons in mafic arc magmas yield ages significantly older than those obtained from Ra-Th mineral isochrons from the same samples. A crystal size distribution case study from Soufrière, St. Vincent, indicates that this may result from the abundance of cumulate xenocrysts within more mafic arc magmas. (b) The majority of U-Th mineral isochrons in intermediate arc magmas yield ages within error of eruption age, indicating that other dating techniques that may be more suitable for deciphering the shorter timescales yielded by these compositions. In a case study of crystals of remobilized andesites from the Soufrière Hills volcano, Montserrat, intra-crystalline trace element diffusion modeling yields crystallization timescales of the order of 10-1000 years. Varying temperature-time histories of individual crystals can be interpreted to reflect multi-stage magma ascent and the development of ephemeral subvolcanic magma reservoirs. (c) U-Th mineral isochrons from felsic arc magmas yield very variable ages from >200 ka to shortly prior to eruption. A case study from the Taupo Volcanic Zone has shown that both remobilization of previous intrusives and melting of old country rock are occurring, giving rise to a this range in ages. Both intermediate and felsic magmas therefore can be shown to have multi-stage ascent histories. Future work will have to address the reasons for the apparent inter-volcano variability in time-averaged subvolcanic magma volumes within the arc crust. These include differences in magma supply rate; thickness, composition and thermal structure of the crust; and magma composition and volatile content, which may vary between different tectonic settings.