During the last decade, solution structures of many small proteins have been solved by NMR. The size of proteins that are being analyzed by NMR seems to increase steadily. Protein structures up to 18 kD have been solved so far, and spectra of proteins up to 30 kD have been assigned. Thus, NMR emerges as an attractive technique, in particular for structural studies of proteins that cannot by crystallized. However, the application of the technology is limited by relaxation properties of the proteins. If relaxation would only be determined by Stokes-Einstein-type rotational diffusion, the effects of the molecular size on relaxation properties of proteins and thus on the performance of multi-dimensional multiple-resonance experiments could readily be estimated. From this perspective, solving two- or three-fold larger structures seems possible. However, most larger proteins exhibit serious line broadening due to aggregation or other still unknown effects. Sample conditioning to minimize these effects is presently the challenge in the work with large proteins.