Studies of magmatic systems have long used the textures of erupted samples to infer processes that control the location and duration of magma storage and drive volcanic eruptions from these storage regions. Models of volcanic processes and magmatic systems have evolved substantially over the past decades, in large part because of advances in analytical and experimental techniques. Cooling- and decompression-experiments have greatly enhanced our understanding of crystal textures produced by crystallization associated with volcanic eruptions, while advances in compositional mapping, isotopic analysis and diffusion chronometry provide the tools to unravel complex histories of individual crystals. Experiments, however, have failed to replicate the full range of groundmass textures observed in volcanic samples and the recognition that magma commonly includes both indigenous (grown from the transporting liquid) and exogenous (incorporated from elsewhere in the system) crystals complicates interpretation of crystal populations in volcanic samples. Analysis and interpretation of crystal size distributions (CSDs) and other physical measures of crystal populations, in particular, have yet to fully account for crystal populations with diverse origins and growth histories. Here I assess the extent to which experiments replicate observed crystal populations and thus can be used to improve understanding of volcanic processes. I then review conditions under which the size characteristics of crystal populations can be reasonably interpreted, examine possible reasons for experimental failure to achieve the very high crystal number densities that characterize some eruptive samples, and suggest ways to link CSD analysis to other techniques that seek to constrain the origin of the complex crystal populations.