Narrow proton signals, high sensitivity, and efficient coherence transfersprovided by fast magic-angle spinning at high magnetic fields make automated projection spectroscopy feasible in protein solid-state NMR. We present the first ultra-high dimensional implementation of this approach where 5D peak lists are reconstructed from a number of 2D projections for protein samples of different molecular size and aggregation state, featuring limited dispersion of chemical shifts or inhomogeneous broadenings. The resulting datasets are particularly suitable to automated analysis, yielding rapid and unbiased backbone resonance assignments.