The first generations of stars are thought to have been more massive than Pop I stars and therefore some of these are thought to have produced pair creation supernovae (PCSNe) at the end of their life. However, the chemical signature of PCSNe is not observed in extremely metal poor stars (e.g. Umeda and Nomoto 2002) and it raises the following questions: Were stars born less (or more massive) than the mass range expected to lead to the PCSNe? Or was mass loss too strong during the evolution of these stars and prevented them from retaining enough mass to produce PCSNe? The discovery of very massive stars (VMS, M > 100 M⊙) in the Milky Way and LMC (Crowther et al. 2010) shows that VMS can form and exist. The observations of PCSN candidates (2006gy & 2007bi) also seems to indicate that such SNe may occur. Mass loss plays a crucial role in the life of VMS since the star will only die as a PCSN if the star retains a high mass throughout its life. In this paper, we shall describe the dependence of VMS evolution on metallicity and present stellar evolution models at various metallicities, including the effects of mass loss and rotation. Based on our models, we will give our predictions concerning the fate of these VMS, either a PCSN or SNIc (possibly GRBs in some cases) as a function of metallicity.