The spin–lattice relaxation times of protons in simple organic molecules are shown to fit a modification of Stokes's model for rotational and translational diffusion in liquids if it is assumed that only dipolar interactions are important in causing relaxation. If the Stokes frictional coefficients for diffusion of a sphere are modified by a “microviscosity factor” which allows for the discontinuous nature of the surrounding fluid, quite good agreement is obtained between the calculated and observed relaxation times, and it is therefore possible to predict approximately relaxation times for protons in different structures.