We have measured the ESR linewidths of ClO2 as a function of temperature in a number of solvents. The dominant spin relaxation mechanism is the motional modulation of the spin–rotational interaction; therefore, the measurements yield the autocorrelation times for angular momentum. The experimental data can be correlated with the aid of the relation τJα−1 = (8πr03ηκ / Iα) + Δα2[τJβτJγ / (τJβ + τJγ)], where τJα is the autocorrelation time for the component of angular momentum along the αth molecular axis (α = x, y,z), Iα is the moment of inertia about the αth axis, η is the coefficient of viscosity for the solvent, r0 is a translational hydrodynamic radius which is determined experimentally by means of diffusion experiments, Δα is the frequency of precession of the αth component of angular momentum, a frequency which vanishes for spherical tops and linear molecules and can be calculated entirely in terms of known components of the moment of inertia, and κ is an adjustable parameter which is assumed to be independent of temperature and viscosity but dependent upon solvent. There are three such equations for the three correlation times τJα, and they can be combined with linewidth formulas to yield the ESR linewidth as a function of T, η, and κ. The linewidth-versus-(T / η) curves can then be fitted by adjusting the value of the single parameter κ for each solvent. The parameter κ1/2 is a measure of the anisotropy of the intermolecular potentials; it approaches unity for very anisotropic interactions and zero for hard-sphere interactions. A derivation of the equation above is given along with an analysis of the relation between κ and the intermolecular potential. Various corrections to the results are presented together with a critique of the present theory and a comparison with other work.