The behaviour of beams and torsion specimens when used in stress-relaxation experiments is analysed for some ideal solids with simple creep laws. It is shown how the experimentally observable deformations can be forecast and the difficulties in the reverse process, deriving a creep law from the experimental observations, are made obvious. Possible errors in the analysis of some irradiation creep tests are discussed, and it is shown that to one set of experimental data at least two models can be fitted equally satisfactorily.