Abstract: Ductile fracture by nucleation, growth and coalescence of internal voids is the dominant fracture mechanism in metals at ambient temperature. Micromechanics-based models for each elementary mechanism have been developed and enhanced over the past 40 years, allowing microstructure-informed failure predictions essentially assuming homogeneous damage evolution. In situ 3-D microtomography has been instrumental to assess these models from experimental evolution of average void density and size. Nevertheless, statistical effects on the damage evolution associated with microstructure distribution heterogeneities have not yet been addressed quantitatively due to the difficulty in tracking individual cavities. Here, we show, from 3-D in situ microtomography characterization of a Ti–6Al–4V alloy, factor 4 variations among the growth rate of individual cavities undergoing the same stress triaxiality and same plastic deformation. This statistical analysis has been made possible owing to an advanced tracking algorithm relying on a graph-based data association approach initially developed for the field of computer vision to track target motions. The measured variations originate from void shape and crystal orientation effects, as well as from local constraints changes due to the presence of two phases with different strengths.