We have coupled three independent methods to investigate the time evolution of eruptive intensity during the sub-Plinian andPlinian phases of the 3600-year BP Late Bronze-Age eruption of Santorini Volcano: (1) mass eruption rate based on new lithicisopleth maps for multiple layers of the fall deposit, (2) magma decompression rate calculated from vesicle number densities, and(3) magma decompression rate calculated from H2O gradients in melt reentrants, with methods 2 and 3 measured on the samesuite of pyroclasts. Mass eruption rate increased by two orders of magnitude, reaching 210 × 106kg s−1at the peak of the Plinianphase (plume height 28.4 ± 1.0 km); it then declined in the final stage of fallout emplacement following the first generation ofpyroclastic surges. Decompression rates from melt reentrants (0.008 to 0.25 MPa s−1) are two to three orders of magnitude lowerthan those from vesicle number densities, assuming heterogeneous vesicle nucleation (2 to 19 MPa s−1). Melt reentrants arethought to record slow decompression in the deep feeder conduit, whereas vesicles record much higher rates of decompression inthe shallow conduit due to the steep, nonlinear pressure gradients associated with magma vesiculation and fragmentation.Upwardly converging flow from a dike-like, deep conduit to a more cylindrical, shallow conduit may also have played a rolein causing upwardly accelerating flow. Variations in deep decompression rate recorded by melt reentrants are decoupled frommass eruption rate, whereas those recorded by vesicles lie in between. Taken with the transition from unsteady to steady Plinianeruption, this may reflect the existence of transient flow conditions in the conduit system due to widening and lengthening of adeep feeder dike as Plinian eruption progressed. As the mass eruption rate rose to its peak value, the fragmentation level fell in theconduit due to increasing rates of magma strain and decompression.