Photochromic fluorescent proteins play key roles in super-resolution microscopy and optogenetics. The light-driven structural changes that modulate the fluorescence involve both trans-to-cis isomerization and proton transfer. The mechanism, timescale and relative contribution of chromophore and protein dynamics are currently not well understood. Here, the mechanism of off-to-on-state switching in dronpa is studied using femtosecond-to-millisecond time-resolved infrared spectroscopy and isotope labelling. Chromophore and protein dynamics are shown to occur on multiple timescales, from picoseconds to hundreds of microseconds. Following excitation of the trans chromophore, a ground-state primary product is formed within picoseconds. Surprisingly, the characteristic vibrational spectrum of the neutral cis isomer appears only after several tens of nanoseconds. Further fluctuations in protein structure around the neutral cis chromophore are required to form a new intermediate, which promotes the final proton-transfer reaction. These data illustrate the interplay between chromophore dynamics and the protein environment underlying fluorescent protein photochromism.