This paper reports the results of experiments on projectile impact into regolith targets at various impact angles. Copper projectiles of 240 mg are accelerated to 197 to 272 m s -1 using an electromagnetic gun. The ejecta are detected by thin Al foil targets as secondary targets, and the resulting holes on the foil are measured to derive the spatial distribution of the ejecta. The ejecta that penetrated the foil are concentrated toward the downrange azimuths of impacting projectiles in oblique impacts. In order to investigate the ejecta velocity distribution, the nondimensional volume of ejecta with velocities higher than a given value is calculated from the spatial distribution. In the case of the vertical impact of the projectile, most ejecta have velocities lower than 24% of the projectile speed (βΌ50 m s -1 ), and there are only several ejecta with velocities higher than 72 m s -1 . This result confirms the existence of an upper limit to the ejection velocity in the ejecta velocity distribution (Hartmann cutoff velocity) (W. K. Hartmann, 1985, Icarus 63, 69-98). On the other hand, it is found that, in the oblique impacts, there are a large number of ejecta with velocities higher than the Hartmann cutoff velocity. The relative quantity of ejecta above the Hartmann cutoff velocity increases as the projectile impact angle decreases. Taking these results with the results of S. Yamamoto and A. M. Nakamura (1997, Icarus 128, 160-170) from impact experiments using an impact angle of 30 β’ , it can be concluded that the ejecta from these regolith targets exhibit a bimodal velocity distribution. Below a few tens of m s -1 , we see the expected velocity distribution of ejecta, but above this velocity we see a separate group of high-velocity ejecta.