Magnetic particles are very efficient magnetic resonance imaging (MRI) contrast agents. In recent years, chemists have unleashed their imagination to design multi-functional nanoprobes for biomedical applications including MRI contrast enhancement. This study is focused on the direct relationship between the size and magnetization of the particles and their nuclear magnetic resonance relaxation properties, which condition their efficiency. Experimental relaxation results with maghemite particles exhibiting a wide range of sizes and magnetizations are compared to previously published data and to well-established relaxation theories with a good agreement. This allows deriving the experimental master curve of the transverse relaxivity versus particle size and to predict the MRI contrast efficiency of any type of magnetic nanoparticles. This prediction only requires the knowledge of the size of the particles impermeable to water protons and the saturation magnetization of the corresponding volume. To predict the T2 relaxation efficiency of magnetic single crystals, the crystal size and magnetization – obtained through a single Langevin fit of a magnetization curve – is the only information needed. For contrast agents made of several magnetic cores assembled into various geometries (dilute fractal aggregates, dense spherical clusters, core–shell micelles, hollow vesicles…), one needs to know a third parameter, namely the intra-aggregate volume fraction occupied by the magnetic materials relatively to the whole (hydrodynamic) sphere. Finally a calculation of the maximum achievable relaxation effect – and the size needed to reach this maximum – is performed for different cases: maghemite single crystals and dense clusters, core-shell particles (oxide layer around a metallic core) and zinc-manganese ferrite crystals.