Nuclear magnetic resonance imaging can reach microscopic resolution, as was noted many years ago, but the first serious attempt to explore the limits of the possibilities was made by Hedges. Resolution is ultimately limited under most circumstances by the signal-to-noise ratio, which is greater for small radio receiver coils, high magnetic fields and long observation times. The strongest signals in biological applications are obtained from water protons; for the usual magnetic fields used in NMR experiments (2-14 tesla), receiver coils of one to several millimeters in diameter, and observation times of a number of minutes, the volume resolution will be limited to a few hundred or thousand cubic micrometers. The proportions of voxels may be freely chosen within wide limits by varying the details of the imaging procedure. For isotropic resolution, therefore, objects of the order of (10μm) may be distinguished.Because the spatial coordinates are encoded by magnetic field gradients, the NMR resonance frequency differences, which determine the potential spatial resolution, may be made very large. As noted above, however, the corresponding volumes may become too small to give useful signal-to-noise ratios. In the presence of magnetic field gradients there will also be a loss of signal strength and resolution because molecular diffusion causes the coherence of the NMR signal to decay more rapidly than it otherwise would. This phenomenon is especially important in microscopic imaging.