Low‐velocity impact experiment on rocks was performed to reveal a nature of collision process of planetesimals. Projectiles of mild steel (S15CK) and rocks (tuff and basalt) with velocities of 17 to 270 m/s were impacted against four kinds of rocks (tuff, basalt, granite, and dunite) with various shapes and sizes. Imparted energy divided by target mass Ei ranges 106 to 108 erg/g. The phenomena associated with the impact of mild steel against spherical targets were classified into five categories. The categories vary with increase in impact velocity as follows: (1) rebound of projectile with no contact damage on target, (2) rebound of projectile with fragmentary chips, and creations of a contact damage (crater?) and radiating fissures on target, (3) rebound of projectile with longitudinal splitting of target, (4) destruction of target with producing shatter conelike fragments, and (5) complete destruction of target. We cannot see any difference in the above categories between various kinds of rocks. The shape of the target also does not influence the categories significantly. The size distribution of the fragments produced by the complete destruction was well expressed by an inverse power law relation, n(l) dl ∝ l−α−1dl, where n(l) is the incremental number of the fragments within a linear increment dl, l is the size, and α is constant; α seems to increase with increase in Ei,. An empirical relationship between α and Ei was approximately given by α∝ log Ei. Empirical formulae for the maximum fragment mass normalized by the original target mass (Ml/Mt) and Ei were also presented for each rock type. It is found from the shape distribution of the fragments produced by the shatter cone type destruction that the length ratios between the shortest and the longest axes, c/a, and between the intermediate and the longest axes, b/a, were greater than 0.2 and 0.3, respectively. Average length ratio of a:b:c is roughly given by 2:√2:1. Collisional phenomena associated with low‐velocity impact experiment were shown to be different from those observed in the high‐velocity impact experiment. The difference in mechanical properties between projectile and target significantly influenced the collisional behavior. For collision between rocks with similar mechanical properties, relative size between projectile and target played an important role in determining the collisional mode: Only the smaller body was destroyed completely, and the larger body suffered no damage even if released Ei was enough to destroy both colliding bodies simultaneously. This observation suggests an important implication in the early stage of planetary formation, that runaway growth of the largest planetesimal takes place even when the collision condition prefers catastrophic destruction to other collision types.