When skiing, excessive ski vibrations can affect comfort, control and performance. Through the years, much efforts have been invested in understanding and reducing these ski vibrations. However, the vibration response of skis is still poorly understood as most laboratory measurements do not correlate with the on-snow response. This paper aims to understand the differences between ski vibrations measured in a laboratory setting and the actual in-use response. To that end, a miniature custom data-logging system was developed to measure the on-snow vibration response of an alpine ski during a turn and its associated operational deflection shapes. The operational deflection shapes of two fore-body resonant frequencies (22 Hz and 66 Hz) and one aft-body resonant frequency (37 Hz) were measured using that system and their frequencies agree with previously documented on-snow measurements. These operational deflection shapes also show that the ski's first three resonances are coupled in bending and torsion, and that the ski fore-body and aft-body are coupled in torsion. New tools were also developed to facilitate quantitative vibration testing of skis, including a custom test bench for controlled skidding and a ski simulation that include a simple snow interaction model. The operational deflection shapes obtained with these tools showed differences in resonance frequencies and amplitudes. Nevertheless, the results showed good overall agreement with on-snow measurements and confirmed that the observed vibration response and deflection shapes were the result of coupling between bending and torsion caused by the ski being on edge.