In the literature a good reliability of isokinetic measurements is reported. In spite of these reports high variations of the peak torque and work values were registered when testing 24 patients with low back pain, using a CYBEX TR isokinetic trunk rotation apparatus. After a 3 week's training the patients force (peak torque) increased in the average between 10.7 p.c. (SD +/- 26.4) and 23.2 p.c. (SD +/- 63.8), corresponding to the test-velocity. Seven of the 24 patients improved or deteriorated more than 30 p.c. Analysing the parameters which could interfere with the results there was found a discrepancy between the original CYBEX software and the German Iso-Kin 4.0 software. The Isokin 4.0 does not warm up and calibrate the isokinetic apparatus daily. Investigations with definite weights showed that the values for peak torque are 40-50 m higher when using the Iso-Kin 4.0 software in comparison to the original CYBEX software. Furthermore there are a lot of other parameters such as acceleration of the body, psychological alterations, pain, motivation, learning effects, and artificial oscillations of the registered curve, which can affect the results even when using the original CYBEX software.