The necessity of atom movements for both creep and self-diffusion suggests a method of correlating the constant-stress creep properties of pure metals. The concept of steady-state creep is discarded, and two empirical approximations for the strain-time behavior of pure polycrystalline metals lead to a creep equation defining an activation energy as the only undetermined constant. For cases where the orientation of individual crystals is important, a second constant is required. Application of this equation to published creep data effects a correlation which indicates that the apparent activation energies observed for creep and self-diffusion show a similar temperature dependence. The effect of stress upon the activation energy for several metals is described approximately, but the need for further experiments encompassing lower values of stress is revealed. The qualitative effects of impurities, grain size, cold-working, and surface conditions upon creep as predicted by a diffusion analogy are found to be in agreement with experimental results, but it is noted that the analogy does not hold if creep deformation is obtained as a result of slip.