This study examines transient creep of single crystals of both natural and synthetic iron-bearing olivine under uniaxial compression (0.1 MPa confining pressure and loads of 25-30 MPa) at high temperature (1650 K) and controlled oxygen fugacity. Natural samples were obtained from San Carlos, Arizona and synthetic crystals were grown at Lawrence Livermore National Laboratory. Samples were deformed in the c and c orientations, corresponding to the softest and intermediate strength orientations, respectively, as determined from steady-state creep tests. Dislocation microstructures were examined for samples unloaded at 0%, 0.1%, 0.5% and 5% strain. At 0% strain the dislocation density and morphology were lower and less complex in the synthetic olivine than in San Carlos samples. Nearly all microstructures initially present in undeformed material were overwritten by 0.1% strain. With further straining, little change in microstructure occurred to the 5% strain limit tested here. Dislocation microstructures in c samples were consistent with the activation of the (001) and (100) slip systems. Microstruc- tures formed in c samples matched those expected from activation of a single slip system, (010). These slip systems are the same as those identified as responsible for steady-state creep under similar temperature, oxygen fugacity and stress conditions. Both natural and synthetic crystals deformed under constant stress in c showed normal strain hardening with initial strain rates about an order of magnitude higher than those at 5% strain. After an initial high strain rate that was roughly equal in both sample types, the synthetic samples deformed at higher rates than the natural crystals. Crystals in the c orientation deformed in a strikingly different manner. San Carlos olivine showed inverse or strain-softening creep in the first 1-2% strain, after which there is weak evidence suggesting a change to strain-hardening behavior. The temporal strain behavior of synthetic olivine in c is strongly sigmoidal. An inflection point at 1% strain marks the change from inverse to normal transient creep. In the c orientation, steady-state creep was not attained for synthetic samples in the 5% strain limit tested here. These results imply that anisotropic transient creep may exist in the upper mantle, complicating rheological models of post-glacial rebound. The transient creep observed in the c orientation illustrates that the strain- hardening Burgers body model is not universally applicable. The (100) slip system is not always the softest system in the transient regime. Strain may initially be accommodated primarily in the (001), (100) duplex system. Finally, the transient regime has been shown to extend to several percent strain (in the c. orientation), making transient creep potentially important in modelling initial post-glacial rebound.