The stress and temperature dependence of the activation volume is often explained by a similar dependence of the measured stress sensitivity. This explanation cannot be reconciled with the nonconservative motion of jogs, with dislocation climb, nor with the intersection mechanism. It is in contradiction with the results obtained in direct dislocation velocity measurements as well. It is now shown that when the backward movement of the dislocations is taken into consideration an explanation, consistent with the rate theory and with the dislocation mobility observations, can be developed. The analysis shows that with constant activation volume the stress sensitivity is stress and temperature dependent, in agreement with the corresponding measurements and mechanisms.