An image pickup apparatus including a lens barrel with a bending optical system, in which a reflection optical element is movable between a storage state and a photographing state. When a rotary cylinder that retains a lens group to which light beams from an object are incident is being driven by a drive source to advance and retreat between a storage state and a photographing state, the coupling between a cam mechanism for rotary cylinder operation and a driving force transmission gear train for optical element operation is automatically established and released, whereby the operation for moving the reflection optical element between the storage state and the photographing state can be achieved by a single drive source.