This research report presents a bundle adjustment technique which is independent on camera intrinsic parameters. Standard bundle adjustment techniques consider camera intrinsic parameters as unknowns in the optimization process. The scheme proposed in this report differs from previous standard techniques since unknown camera intrinsic parameters are computed as a function of the 3D structure and the camera pose. Considering less unknowns in the optimization process produces a faster algorithm which is more adapted to real-time applications. Computationally expensive metric reconstruction, using for example several zooming cameras, considerably benefits from an intrinsics-free bundle adjustment. Indeed, experimental results, obtained using both simulated data and real images, show that the algorithm we propose not only considerably reduces the computational cost but also improves the accuracy of the metric reconstruction.