The electron spin-lattice relaxation and spectral diffusion properties at 22 K have been examined for three tyrosine radicals: the stable tyrosine radical YD+ in photosystem II (PSII), the tyrosine radical in the B2 subunit of ribonucleotide reductase from Escherichia coli, and a model L-tyrosine radical prepared in a frozen glass by UV irradiation at 77 K. The relaxation transients obtained by the method of saturation recovery, where recovery from saturation following a microwave pulse is detected by a low continuous observation microwave level, were compared to those obtained by the inversion-recovery and spin-echo-detected saturation-recovery techniques. The same values for the electron spin-lattice relaxation rate 1 T1 were obtained by saturation recovery and spin-echo-detected saturation recovery provided that spectral diffusion channels were saturated by a sufficiently long microwave pulse. The inversion-recovery spin-echo method produced relaxation transients that were more rapidly decaying than those of either of the saturation-recovery methods, owing to the presence of spectral diffusion. Two contributions to spectral diffusion were observed. For the UV-generated model L-tyrosine radical, the dominant contribution to spectral diffusion occurred on the time scale of the electron spin-spin relaxation time T2 and was saturated with a microwave pulse lasting only 50 μs. The spectral diffusion processes for the tyrosine radicals in the proteins occurred on the time scale of T1; a saturating microwave pulse of duration greater than 3-5 ms was required to completely saturate these spectral diffusion processes. These slow spectral diffusion processes are likely to involve electron-nuclear spin-spin interactions. It is probable that the large separation between radicals imposed by the dimensions of the protein prevents spectral diffusion paths due to interactions between tyrosine radicals on the time scale of T2. The UV-generated l-tyrosine radical and the tyrosine radical in ribonucleotide reductase exhibited predominantly single-exponential relaxation kinetics when very long saturating pulses were employed. The tyrosine radical in PSII, however, exhibited nonexponential relaxation kinetics even at the extrapolated limit of infinitely long saturating pulses, where contributions from spectral diffusion have been eliminated. We suggest that spin-spin interactions of the tyrosine radical with neighboring paramagnetic site (s) are the source of the deviation from exponential relaxation kinetics in PSII.