Assuming that the equilibrium geometry of river channels does not depend on their initial state but solely on boundary conditions, several formulas have been derived that relate the channel depth and width to the river bankfull discharge and bed material. However, due to the existence of a threshold for sediment motion and the strong non-linearity between sediment transport and flow rate, this assumption might not be generally valid for gravel-bed rivers. This research clarifies the role of the initial conditions, more specifically the initial channel width, on the geometry of gravel-bed rivers considering a variety of boundary conditions. The approach includes laboratory experiments and two-dimensional modeling, reproducing the evolution of alluvial channels with different starting widths, discharge regimes and sediment input rates. The experiments represent the Arc River (France). Thus, the characteristics of this river were used in the numerical model to obtain a realistic virtual case complementing the experiments. Different boundary and starting conditions resulted in either braided or single-thread channels. We found that the initial width strongly influences the evolution process and leaves a footprint on the river braid-belt extension. The active width of braided systems and the width of single-thread channels do not depend on the starting condition. They depend on sediment input rather than on discharge variability. Different initial widths result in different final bed levels. This indicates that the initial channel width may affect the degree of channel incision or aggradation. The results of this study justify the use of equilibrium formulas for single-thread rivers.