Abstract New steady-state creep data for a dunite and a peridotite at 1000°C and 6 kbar pressure are presented and compared with published studies of similar rocks at 5–15 kbar. Shear stresses here ranged from 1 to 3 kbar. A current best estimate of the flow law for dunite at 1000°C is presented, although some major inconsistencies exist in published high-pressure work. Our current understanding of the mechanisms involved do not reasonably allow an extrapolation of this data to geologic strain rates. Relatively little extrapolation is required, however, to show that laboratory strain rates appear to be significantly higher than those indicated by geophysical evidence. When the effects of possible partial melting are taken into account, this discrepancy is increased. New results are given for transient creep in diabase, gabbro, and granite close to the solidus and peridotite at 1000°C. For strains up to about 10 −3, strain closely follows a power law with time with an exponent ranging from 0.35 to 0.50, depending on rock type. Compared with calculations based on glacial rebound, the transient creep we observe near the solidus would give too rapid recovery. Thus, for both large and small strains, our materials are considerably weaker at the solidus than is required by geophysical evidence.