The hypothesis of variables separation, namely the time and the strain separation in the relaxation function, is widely used in soft tissue biomechanics. Although this hypothesis is central to several biomechanical models, only few experimental works have tried to verify it. From these studies, contradictory results have been found. Moreover, it has recently been noted that no such experimental verification has been performed for ligament tissues. In this paper, an experimental method is developed to test the hypothesis of variables separation. This method is then used with human cruciate ligaments and patellar tendons. It is shown that the use of the variables separation hypothesis is justified at least for strain values lower than 16% for anterior cruciate ligament, lower than 12% for posterior cruciate ligament and lower than 6% for patellar tendon. The method presented in this paper could be used to verify the validity of variables separation for other tissues.