Biological function largely depends on protein dynamics. Protein-protein interactions and ligand recognition have been shown to occur on a µs time scale. 1 In membrane proteins, the presence of slow motion has been a hurdle for X-ray crystallography, as in the case of the 2 -adrenergic G-protein coupled receptor (GPCR). 2 Also for NMR spectroscopy, intermediate (ns-µs) motion tends to be an obstacle. In many amyloidogenic peptides and proteins, large parts of the primary sequence are often obscured and do not yield detectable resonances in solid-state NMR spectra, presumably due to dynamics. 3 In the voltage gated membrane protein VDAC, 4,5 large segments of the constricting N-terminal R-helix, which seems important for the gating process, could not be assigned due to dynamics. 5 Ironically, those parts of a protein undergoing slow motion are particularly interesting for the understanding of its biology. We and others have demonstrated that sparsely protonated proteins can be successfully employed in Magic Angle Spinning (MAS) solid-state NMR for proton detection, 6 INEPT based resonance assignment strategies, 7 and mapping of solvent accessibility. 8 This approach allowed us to show for the first time that mutual cancellation of dipole-dipole and chemical shift anisotropy (CSA) relaxation pathways 9,10 can give rise to differential transverse relaxation times T 2 also in the solid state. While in the solution state, the effect is due to isotropic tumbling of the molecule, differential relaxation in the solid-state was shown to be due to internal mobility. 11,12 We show here that TROSY 10 type scalar coupling based triple resonance experiments in combination with perdeuteration are beneficial for the detection and assignment of those residues in the protein which undergo intermediate (ns-µs) dynamics. Standard (non-spin-state selective) triple resonance experiments, in contrast, perform extremely poorly in these cases. 7 Conventional MAS solidstate NMR experiments that are based on dipolar transfers are shown not to detect these flexible residues at all. Figures 1A and B depict the first 1 H/ 15 N planes of tripleresonance out and back HNCO experiments, in which 1 H magnetization is transferred to C′ and back. For these experiments, a sample of the chicken R-spectrin SH3 is employed which is 100% perdeuterated at nonexchangeable sites and partially back-exchanged with protons at labile sites. 7 Paramagnetic Relaxation Enhancement (PRE) using Cu-edta was employed for accelerated data acquisition. 13 While the spectrum in Figure 1A was recorded using a standard HNCO pulse scheme, 14 the spectrum in Figure 1B was recorded with spin-state selection, 15 using the pulse schemes shown in Supplementary Figure 1E and F, respectively. Even for the temperature which yields the best signal-to-noise ratio, residues located in mobile parts of the protein (labeled in black) have severely reduced intensities in comparison to residues that are situated in immobile -sheet regions (labeled in gray). For those dynamic residues, sel...