The practitioner interested in reducing software verification effort may found herself lost in the many alternative definitions of Graphical User Interface (GUI) testing that exist and their relation to the notion of system testing. One result of these many definitions is that one may end up testi
ng twice the same parts of the Software Under Test (SUT), specifically the application logic code. To clarify two important testing activities
for the avoidance of duplicate testing effort, this paper studies possible differences between GUI testing and system testing experimentally. Specifically, we selected a SUT equipped with system tests that directly exercise the application code; We used GUITAR, a well-known GUI testing software to GUI test this SUT. Experimental results show important differences between system testing and GUI testing in terms of structural coverage and test cost.