In Nuclear Magnetic Resonance (NMR) spectroscopy, the measurement of the collective spin magnetization is weakly invasive and its back-action is called radiation damping. The aim of this paper is to provide a control-theoretical analysis of the problem of suppressing radiation damping effects. We show that the various real-time feedback schemes commonly used in NMR can be cast in terms of high gain feedback, of exact cancellation based on knowledge of the radiation damping field, and of 2-degree of freedom control designs, with the exact cancellation as prefeedback. We further show that the formulation in control-theoretical terms naturally leads to devising other possible closed-loop schemes, such as a general high gain feedback stabilization design not requiring the knowledge of the radiation damping field.