The work presented in this paper comes as part of a research program dealing with the thermomechanical behaviour of rock salt. It aims to study laboratory and in-situ long-term behaviour by means of creep tests with deviator and temperature changes. The laboratory results, using a triaxial multi-stages creep tests, highlighted the strain hardening character of rock salt. Furthermore, the in-situ results, using a borehole dilatometer multi-step creep test, have shown that the drilling is carried out in a weakly stressed pillar. The interpretation of the laboratory results, using the J.LEMAITRE law, did not indicate full agreement with all the test results. As a result a 'double' J.LEMAITRE model, which takes into account a double strain hardening variable, has been put forward. The validation of this model on the laboratory creep tests is very satisfactory. Furthermore, the activation energy seems satisfactory to represent the influence of the temperature. The in-situ behaviour modelling is clearly more complex than the modelization based on laboratory tests. In fact, it seems that if the rock salt behaviour is maintained by J.LEMAITRE law, it is necessary to vary with the stress, at least, one of the parameters assumed constant in the basic law.