It is common to employ a traditional double cell system, of which an open-ended inner cell is installed in an ordinary triaxial apparatus, to measure the volume change of unsaturated specimens. In such a system, the total apparent volumetric strain of the specimen (ev) is deduced from the water level change in the inner cell, monitored by a differential pressure transducer (DPT) considering, meanwhile, the top cap intrusion into the inner cell recorded by a vertical displacement transducer (VDT). Severe apparent volumetric strain, caused by the compliance of the double cell system(ev,SC), was observed during the undrained cyclic loading tests in a previous study. Test results on a steel-spring dummy specimen revealed that ev, SC was induced not only by such as themeniscus effect, but unexpectedly also by the asynchronous responses between the DPT and the VDT (i.e., the response of the DPT was delayed compared with that of the VDT). By doing some treatment ev, SC could be reduced to some extent, whereas the magnitude of ev, SC was still too high to be acceptable when the tested specimen approached the liquefied state. To radically solve these technical difficulties, a modified double cell system, named the linkage double cell system, was developed in this study. In this modified system, a linkage rod moving simultaneously with the loading shaft was introduced, through which the DPT could directly measure ev without considering the top cap or loading shaft intrusion. Test results for the steel-spring dummy specimen as well as for saturated and unsaturated soil specimens demonstrated that the linkage double cell system has major advantages in measuring accurately the volume change of the specimen during undrained cyclic triaxial loading tests compared with the traditional double cell system.