A prerequisite in refining volcanic hazard at explosive volcanoes is a better quantification of the timescales of the syn-eruptive processes, such as magma degassing and crystallization prior to eruption. To this aim, new data on the matrices (microlites, residual glasses, and bubbles) of pumice, scoria, and dense clasts erupted during the AD 1530 andesitic eruption of La Soufrière of Guadeloupe are combined with published data from phase-equilibrium and kinetic experiments, in order to estimate pressures of microlite crystallization and magma ascent rates. From the timescale data, we infer that the AD 1530 eruption started with phreatomagmatic explosions tapping magmas that ascended during about 1 month (decompression rate of ∼50 Pa/s) from the coldest parts of the reservoir (∼825°C and a 74 wt% SiO 2 melt). These magmas continuously crystallized microlites (∼25 vol% plagioclase, orthopyroxene, clinopyroxene, magnetite, quartz, and apatite), but did not outgas likely due to sealed conduit margins. The flank collapse (preexisting “cold” edifice) that followed the phreatomagmatic phase triggered a sub-Plinian eruption that progressively tapped the hotter main reservoir (∼875°C and 71 wt% SiO 2 interstitial melt), emitting banded and homogeneous pumice. The banded pumice did not significantly outgas and mostly lack decompression-induced microlites, suggesting short ascent durations of the order of 0.5–1 day (decompression rates of 1,400–4,000 Pa/s). The following Strombolian phase emitted dark scoria that did not significantly outgas and only crystallized rare microlites, suggesting ascent duration of the order of 2 days (decompression rates of ∼550 Pa/s). The terminal lava dome growth involved fully outgassed magmas ascended during more than 1 month, giving time for microlite crystallization (∼40 vol% plagioclase, orthopyroxene, clinopyroxene, magnetite, and cristobalite). The detection of any shallow new magmatic intrusion is crucial, since it can trigger a sequence of conduit processes leading to an eruption marked by a succession of different and disastrous eruptive styles, following a scenario similar to the AD 1530 eruption. Overall, we provide a combined approach of petrological, geochemical, and experimental data that may be used to infer ascent conditions and rates at other volcanic systems.