Electron spin lattice relaxation rates (1/T-1e) of nitroxyl radicals for tumbling correlation times between about 0.1 and 10 ns in water:glycerol or water:sorbitol mixtures at room temperature were measured by saturation recovery at X-band (9.2 GHz), S-band (3.1 GHz), and L-band (1.9 GHz) for natural abundance 2,2,6,6-tetramethylpiperidinyl-1-oxy (tempol), tempol-d(17), and N-15-tempol-d(17) and for a spin-labeled derivative of ethylenediaminetetraacetic acid. Tumbling correlation times were calculated from the continuous wave EPR line shapes at X-band. The dependence of T-1e on tumbling correlation time was modeled with contributions from modulation of nitrogen nuclear hyperfine and g anisotropy, from spin rotation and from one or more thermally activated processes. At these microwave frequencies, modulation of nitrogen hyperfine anisotropy makes a substantially larger contribution than modulation of g anisotropy. Spin rotation makes a very small contribution to T-1e at the tumbling correlation times examined in these studies. Replacement of N-14 by N-15 decreases the relaxation rates as predicted for samples in which modulation of nuclear hyperfine anisotropy contributes to relaxation. Deuteration of the solvent does not affect T-1e, which indicates that the electron-nuclear dipolar interaction with solvent nuclei does not make a significant contribution to Tie of nitroxyl radicals in this motional regime. The contributions to relaxation from one or more thermally activated processes that occur at rates comparable to the microwave frequency dominate the relaxation at tumbling correlation times longer than about 2 ns at X-band and S-band. The contributions from the thermally activated process(es) and from modulation of nitrogen nuclear hyperfine anisotropy increase as the microwave frequency is decreased, and the contribution from hyperfine anisotropy dominates at L-band.