Viscous magnetization (VM) and viscous remanent magnetization (VRM) have been measured, as a function of temperature, between room temperature and the Curie temperature using a suite of well-characterized synthetic and natural multidomain (MD) magnetite samples. Particular attention was given to possible diffusion aftereffects such as dislocation creep (stress relaxation) and disaccommodation (vacancy and ionic reordering) and their contribution to viscous behavior in what has been commonly thought of as a purely thermal fluctuation process. Dislocation creep was examined by measuring viscosity before and after annealing. Annealing was found to reduce the non-log(t) behavior, where t is time. Non-log(t) behavior has been associated with diffusion aftereffects, suggesting that these are a major contributor to viscosity and (de)magnetization processes in MD samples. The positive curvature of the non-log(t) acquisition processes indicates that dislocation creep dominates over disaccommodation. This does not imply that VM and VRM are due solely to dislocation creep, but rather that VRM and VM reflect a number of unrelated temperature-dependent processes, primarily thermal fluctuations and dislocation creep. This is the first time that dislocation creep has been directly identified as contributing to viscosity at temperature. These findings will have particular implications for paleointensity determinations, as on heating a sample, its dislocation structure may relax, giving rise to demagnetizations not associated with thermal fluctuations. This will lead to incorrect intensity estimates. If no heating is performed on a geological specimen, then it is very likely that laboratory timescale stress relaxation processes will have already occurred in situ.