Résumé : In the context of intermediate ballistics, a thorough understanding of the muzzle flow is necessary to optimize the design of muzzle devices and long-range projectiles. However, due to the harsh environment, the extremely short duration, and the transient evolution of the intermediate ballistics phase, both experimental and flow modeling efforts are hampered by the lack of quantitative experimental data. In this study, a Particle Imaging Velocimetry (PIV) experimental set-up is implemented to quantitatively investigate the muzzle flow based on its velocity. This tool has the advantage of obtaining a non-intrusive and whole-field diagnostic, both qualitatively and quantitatively. The first part of the study explains that PIV is suitable for measuring the velocity fields of the muzzle flow. This was achieved by using the naturally present particles in the gas as tracers. We demonstrate that the raw PIV images revealed that the structure of this flow is composed of two regions. The first region is located within the under-expanded jet and consists of spatially dispersed particles. The second region, downstream of the Mach disk, is formed by large structures. We show that cross-correlation and particle tracking velocimetry (PTV) algorithms can determine the flow velocity in both regions. In the second part, we present a quantitative description of the muzzle flow issued from the launch of a subsonic .300 Blackout projectile. The results show that these gases reach a maximum centerline velocity of more than 900 m/s inside the shock bottle. At the barrel exit plane, the gases start to discharge with a velocity close to that of the projectile’s launch velocity, accelerating it. In the third part, we present a detailed comparison between the aforementioned flow and the flow resulting from the launch of a supersonic projectile. Differences and similarities are pronounced and explained. The presented set-up and the description of the whole flow field velocity would serve as valuable improvements toward muzzle devices optimization and numerical code validation.