An image pickup apparatus enabling construction of a small bending optical system with high magnification. A first lens group is movably disposed toward an object on a first optical axis. A second lens group is movably disposed radially inward of the first lens group. The second lens group comprises a drive frame drivingly controlled along the first optical axis and a lens holding frame for holding the second lens group. A prism is disposed on the first optical axis, for bending light incident on the prism to thereby guide the light along a second optical axis. The prism is retracted along the second optical axis in a non-shooting state. In the non-shooting state, the second lens group is retracted and accommodated in a space defined by the drive frame and the lens holding frame and a space occupied by the prism in a shooting state.