Abstract: Understanding the characteristics of fracture mode, fragment distribution and its controlling factor for explosively driven metal cylinders are important topics in applied physics, mechanics, weapon engineering and other fields. However, except for numerical simulation, a simple two-dimensional fragmentation model considering fracture mechanism has not put forward. In this paper, the fragmentation of TA2 titanium cylinders with varying charge were carried out experimentally. Through the analysis of macroscopic fracture and microscopic metallographic for the recovered fragments, the fracture mode and its fragmentation distribution of TA2 metal are discussed. The results show that:(1) The fragmentation of TA2 titanium alloy cylinders is all shear fracture modes under different detonation pressure of 7 \sim 25 GPa, but the mechanism is different, which the fracture controlled by multiple adiabatic shear bands at higher explosion pressure; (2) different from the one-dimensional tensile fragmentation, the fragment mass distribution is modeled as an exponential distribution with \beta = 1 due to the insufficient fragmentation of metal cylinders. The \beta can tend to be smaller when the fragmentation of explosively driven cylinders is more sufficient, which would be closer to the Poisson statistical distribution proposed by Mott and Linfoot; (3) The width distribution of cylinder fragments can be well described by the Rayleigh distribution. The normalized width distribution of fragments under different explosion pressures is similar and presents the characteristic of "quantization", which means a minimum characteristic size exists; (4) The characteristic sizes of TA2 shell fragments are much larger than that predicted by the G-K formula, as it is used to describe the distances between multiple adiabatic shear bands. Many factors affect the explosion of cylindrical shell, and the formation of each fragment is random, but the statistical regularity shows that the explosion process and fragment distribution are still predictable overall. The research provides an important reference for fragmentation characteristics, distribution, and model analysis of metal cylinders.